Project-Modeling Data in the Tidyverse
2023-04-02
Load libraries
Training data – complaints from CFPB (data_complaints_train.csv)
train <- read.csv("data_complaints_train.csv")Testing data – complaints from CFPB
test <- read.csv("data_complaints_test.csv")The goal of your project is to predict the consumer complaint category. This is the “Product” variable in the training data set. You may use any of the other variables to predict with. You should create a report describing how you built your model, how you used cross validation, what you think the expected out of sample error is, and why you made the choices you did. You will also use your prediction model to predict 20 different test cases.
Learn about dataset, perform cleaning
Have a look at the training dataset’s summary:
summary(train)## Product Consumer.complaint.narrative Company
## Length:90975 Length:90975 Length:90975
## Class :character Class :character Class :character
## Mode :character Mode :character Mode :character
## State ZIP.code Submitted.via
## Length:90975 Length:90975 Length:90975
## Class :character Class :character Class :character
## Mode :character Mode :character Mode :charactertrain$Product <- factor(train$Product)
levels(train$Product)## [1] "Credit card or prepaid card" "Mortgage"
## [3] "Student loan" "Vehicle loan or lease"train$State <- factor(train$State)
levels(train$State)## [1] "AA"
## [2] "AE"
## [3] "AK"
## [4] "AL"
## [5] "AP"
## [6] "AR"
## [7] "AS"
## [8] "AZ"
## [9] "CA"
## [10] "CO"
## [11] "CT"
## [12] "DC"
## [13] "DE"
## [14] "FL"
## [15] "FM"
## [16] "GA"
## [17] "GU"
## [18] "HI"
## [19] "IA"
## [20] "ID"
## [21] "IL"
## [22] "IN"
## [23] "KS"
## [24] "KY"
## [25] "LA"
## [26] "MA"
## [27] "MD"
## [28] "ME"
## [29] "MH"
## [30] "MI"
## [31] "MN"
## [32] "MO"
## [33] "MS"
## [34] "MT"
## [35] "NC"
## [36] "ND"
## [37] "NE"
## [38] "NH"
## [39] "NJ"
## [40] "NM"
## [41] "None"
## [42] "NV"
## [43] "NY"
## [44] "OH"
## [45] "OK"
## [46] "OR"
## [47] "PA"
## [48] "PR"
## [49] "RI"
## [50] "SC"
## [51] "SD"
## [52] "TN"
## [53] "TX"
## [54] "UNITED STATES MINOR OUTLYING ISLANDS"
## [55] "UT"
## [56] "VA"
## [57] "VI"
## [58] "VT"
## [59] "WA"
## [60] "WI"
## [61] "WV"
## [62] "WY"train %>% glimpse()## Rows: 90,975
## Columns: 6
## $ Product <fct> Credit card or prepaid card, Mortgage, St…
## $ Consumer.complaint.narrative <chr> "I initially in writing to Chase Bank in …
## $ Company <chr> "JPMORGAN CHASE & CO.", "Ditech Financial…
## $ State <fct> CT, GA, IN, MI, MI, FL, WA, CA, CA, VA, G…
## $ ZIP.code <chr> "064XX", "None", "463XX", "490XX", "480XX…
## $ Submitted.via <chr> "Web", "Web", "Web", "Web", "Web", "Web",…Create a barplot of product category: see how many categories there are.
ggplot(data = train, aes(x = Product, fill = Product)) +
geom_bar() +
labs(x = "Product category", y = "Number of occurrences", fill = "Product category")
Calculate how many NAs there are in each variable.
train %>%
map(is.na) %>%
map(sum)## $Product
## [1] 0
##
## $Consumer.complaint.narrative
## [1] 0
##
## $Company
## [1] 0
##
## $State
## [1] 0
##
## $ZIP.code
## [1] 0
##
## $Submitted.via
## [1] 0There are no NAs.
Create parallel threads, for faster running time.
doParallel::registerDoParallel(cores=0.75*parallel::detectCores())In the training dataset, Consumer.complaint.narrative variable, there are strings like XXXX used to redact some data; we’ll get rid of them we’ll also remove numbers, symbols, and make all text in lower case.
train$Consumer.complaint.narrative <- gsub("XX/", "", train$Consumer.complaint.narrative)
train$Consumer.complaint.narrative <- gsub("X/", "", train$Consumer.complaint.narrative)
train$Consumer.complaint.narrative <- gsub("[XX]", "", train$Consumer.complaint.narrative)
train$Consumer.complaint.narrative <- gsub("[[:punct:]]", "", train$Consumer.complaint.narrative)
train$Consumer.complaint.narrative <- gsub("[0-9]", "", train$Consumer.complaint.narrative)
train$Consumer.complaint.narrative <- str_to_lower(train$Consumer.complaint.narrative)
head(train)## Product
## 1 Credit card or prepaid card
## 2 Mortgage
## 3 Student loan
## 4 Credit card or prepaid card
## 5 Credit card or prepaid card
## 6 Credit card or prepaid card
## Consumer.complaint.narrative
## 1 i initially in writing to chase bank in late about a charge that was unauthorized on from in the amount of after many letters and complaints they still did not do anything yet in they closed my original account and reissued a new account and sent me a letter indicating they moved this charge to fraud and it would be immediately credited however they have failed to do that and almost months later it is still on my statement and im getting charged interest even after this i still use my card and pay for my charges and have never been late however i will not pay for a charge i didnt make and refuse and chase refuses to do anything even after many attempts and letters they write stating so i feel this is last step before i begin a case in against them
## 2 my ex husband and myself had a mobile home home mortgage with back in i went through a horrible divorce as my husband was using drugs and failing to pay the bills which led to my having to divorce him the mortgage was in both of our names we only had the mortgage for one year and they got the trailer bc of being unable to pay debt i was not working at the time as i had just given birth to my first child in she is now of age as this happened years ago \n\nnow after years filled a to the irs and the irs alerted me which caused me an adverse affect on my irs taxes \ni received a l letter from the irs which stated that due to ditech mortgage sending a with taxable income of this could cause me to have to owe the irs irs told me to contact ditech financial to dispute this and request that ditech mortgage rescind the in order so it is not taxable income \n\ni contacted ditech at spoke with a and i told her that i did not want to be recorded but she did it anyways \ni told her that i never settled on an amount of of a debt from a mortgage that i had back in with ditech nor did i ever receive a from them i asked them to rescind the taxable income of as this is a debt from years ago in which i should not be taxed for taxable income of on a debt from years ago however refused to rescind reverse the \nnote i have suffered with my credit from this debt for over ten years and it was removed from my debt bc it has been over the to year period that the debt is lawfully removed from my credit they cant go back years later and in send the irs and myself a form to tax me for a debt from over years ago that was charged off my credit report back in the year this is very unfair and not allowed \nalso i disputing the fact that they kept putting this on and off my credit for over years now and it has negatively affected my credit for way over the year period \npoint being if they had plans of taing me at then they should have taxed me yrs ago before the to year period passed which is now way over if they were going to send a taxable income they should have done this years ago not in i am requesting a complaint submitted as they have unfairly affected my credit past the year period and it is violating my rights to be taed for something in on a debt from the year
## 3 i was a student at from i accumulated of debt in student loans this school flat out lied to me i was under the assumption it would have been less than half of that since i had a lot of credits that transferred not to mention this school promised me job placement which they did not they told me i would be making a lot more than what i do they gave me false information the only thing this school did for me is put me in tons of debt and lied to me i am a single mother of and was so excited to get my degree until i graduated and realize how much they lied to me the school is now shut down can i get these loans forgiven due to being lied to and given false information that basically pressured me into enrolling if i would have known that my student loans would have been this high i would have just stayed at the college where i did my prerequisites i feel like i fool that had done this to me
## 4 it has come to my attention the citi group is actively attempting to interfere with rights guaranteed by the constitution nd amendment by manipulating or denying service to certain entities or transactions that are protected by the constitution i do not expect citi group or any other financial institution to be controlling the nation s activities through such unpatriotic and unamerican behavior i do expect citi group to tend to the business of their business and stay out of the business of social tyranny
## 5 this banks new firearm policies run counter to laws and regulations passed by congress and they infringe and discriminate against an individuals second amendment rights such policies should not be endorsed by our federal government which instead should do business with companies that respect all of our constitutional rights including the second amendment our federal government should take all necessary steps to review and terminate its contract with citibank unless they rescind their guidelines
## 6 i only use my walmart store card to keep from having it cancelled my sons a college student had his card cancelled for non use i made a small purchase in and paid it off made another small purchase in and thought i paid it off online via wlamartcom but that seemed not to work i didnt receive a statment after that probably the mail courieri have an apartment style mailbox system where i live nevertheless so i thought all was good until credit sesame showed a past due reporti was out of town at the time so i immediately searched walmarts payment site and paid the account in full i also noticed that my account had been reduced to a mere just short of my now due payment which included aprox in past due fees so it appears that i maxed out the account by a purchase my credit dropped score points from down to from excellent to a fair rating
## Company State ZIP.code Submitted.via
## 1 JPMORGAN CHASE & CO. CT 064XX Web
## 2 Ditech Financial LLC GA None Web
## 3 Navient Solutions, LLC. IN 463XX Web
## 4 CITIBANK, N.A. MI 490XX Web
## 5 CITIBANK, N.A. MI 480XX Web
## 6 SYNCHRONY FINANCIAL FL 331XX WebCreate a term matrix:
matrix <- DocumentTermMatrix(train$Consumer.complaint.narrative)
inspect(matrix)## <<DocumentTermMatrix (documents: 90975, terms: 73921)>>
## Non-/sparse entries: 9149065/6715813910
## Sparsity : 100%
## Maximal term length: 1963
## Weighting : term frequency (tf)
## Sample :
## Terms
## Docs and for have not that the they this was with
## 11002 104 35 41 69 54 396 61 51 48 38
## 11343 113 70 35 42 239 215 33 36 153 52
## 19650 136 73 20 23 59 254 22 6 16 47
## 25455 183 72 50 18 43 251 0 38 27 19
## 27671 134 53 5 45 55 308 3 43 8 34
## 60769 144 57 33 19 149 269 22 28 61 65
## 64243 191 50 31 59 21 197 39 107 106 50
## 68621 199 93 20 46 54 242 40 45 22 70
## 70389 162 41 34 46 85 188 49 76 29 21
## 73540 159 53 16 10 42 202 17 19 7 21Find the most frequent words in the matrix: words that are met at least 10000 times in text:
?findFreqTerms
findFreqTerms(matrix, lowfreq = 10000)## [1] "about" "account" "after" "amount"
## [5] "and" "anything" "bank" "been"
## [9] "before" "card" "case" "charge"
## [13] "charged" "charges" "chase" "closed"
## [17] "did" "didnt" "even" "for"
## [21] "fraud" "from" "getting" "have"
## [25] "however" "immediately" "interest" "last"
## [29] "late" "later" "letter" "make"
## [33] "many" "months" "never" "new"
## [37] "not" "original" "pay" "sent"
## [41] "statement" "stating" "still" "that"
## [45] "the" "them" "they" "this"
## [49] "use" "was" "will" "would"
## [53] "yet" "also" "asked" "back"
## [57] "being" "both" "but" "complaint"
## [61] "contact" "contacted" "could" "credit"
## [65] "debt" "dispute" "done" "due"
## [69] "financial" "first" "given" "going"
## [73] "got" "had" "has" "her"
## [77] "him" "home" "income" "just"
## [81] "mortgage" "now" "off" "one"
## [85] "only" "order" "our" "over"
## [89] "past" "point" "receive" "received"
## [93] "report" "request" "send" "she"
## [97] "should" "spoke" "stated" "submitted"
## [101] "then" "through" "time" "told"
## [105] "very" "want" "way" "went"
## [109] "were" "which" "with" "year"
## [113] "years" "can" "down" "get"
## [117] "how" "information" "into" "like"
## [121] "loans" "making" "more" "out"
## [125] "put" "since" "student" "than"
## [129] "these" "under" "until" "what"
## [133] "where" "any" "are" "business"
## [137] "citi" "other" "service" "their"
## [141] "all" "its" "take" "another"
## [145] "fees" "full" "his" "keep"
## [149] "made" "mail" "online" "paid"
## [153] "payment" "purchase" "score" "via"
## [157] "work" "applied" "call" "called"
## [161] "check" "fee" "help" "need"
## [165] "problem" "provided" "representative" "response"
## [169] "calls" "company" "days" "end"
## [173] "payments" "phone" "loan" "paying"
## [177] "regarding" "see" "statements" "there"
## [181] "advised" "balance" "find" "money"
## [185] "monthly" "once" "vehicle" "when"
## [189] "who" "why" "you" "department"
## [193] "documents" "dont" "every" "issue"
## [197] "modification" "month" "number" "please"
## [201] "process" "funds" "give" "know"
## [205] "provide" "times" "your" "believe"
## [209] "next" "program" "several" "address"
## [213] "again" "because" "claim" "filed"
## [217] "informed" "someone" "used" "during"
## [221] "found" "without" "rate" "refund"
## [225] "requested" "same" "today" "total"
## [229] "tried" "additional" "day" "close"
## [233] "bill" "car" "reason" "said"
## [237] "date" "fraudulent" "insurance" "property"
## [241] "servicing" "already" "correct" "current"
## [245] "fargo" "foreclosure" "wells" "while"
## [249] "house" "lender" "name" "application"
## [253] "navient" "needed" "plan" "situation"
## [257] "customer" "attached" "denied" "email"
## [261] "well" "closing" "escrow" "took"
## [265] "ive" "does" "some" "capital"
## [269] "file" "two" "different" "accounts"
## [273] "nothing" "cards" "able" "notice"
## [277] "explained" "offer" "supervisor" "trying"
## [281] "within" "each" "person" "sale"Remove words like “and”, “a”, “the”, “I”, “my” etc.: they are among the most frequent but they won’t help to predict the product
stopwords <- get_stopwords()
head(stopwords)## # A tibble: 6 × 2
## word lexicon
## <chr> <chr>
## 1 i snowball
## 2 me snowball
## 3 my snowball
## 4 myself snowball
## 5 we snowball
## 6 our snowball# regular expression: \\bword\\b searches for word separated with spaces from other terms in text
train$Consumer.complaint.narrative <- gsub(paste0('\\b', stopwords$word, '\\b', collapse = '|'), '',
train$Consumer.complaint.narrative )Create new term matrix:
matrix2 <- DocumentTermMatrix(train$Consumer.complaint.narrative)
inspect(matrix2)## <<DocumentTermMatrix (documents: 90975, terms: 73819)>>
## Non-/sparse entries: 7015453/6708668072
## Sparsity : 100%
## Maximal term length: 1963
## Weighting : term frequency (tf)
## Sample :
## Terms
## Docs account bank card credit loan mortgage payment payments received told
## 19650 4 34 0 3 49 10 6 7 12 0
## 25455 79 2 0 11 81 58 29 4 3 0
## 27671 2 1 0 1 4 4 8 5 1 0
## 58908 23 2 0 23 27 0 22 32 3 0
## 64243 0 0 0 0 26 15 1 10 4 2
## 68621 9 0 0 10 56 0 16 9 10 8
## 70389 1 0 0 3 2 26 0 3 0 0
## 73540 8 1 0 6 33 61 7 37 3 0
## 77518 0 0 0 0 837 0 45 1 0 2
## 84551 38 0 0 8 70 60 22 16 0 0Remove infrequent words, to reduce matrix size:
matrix2 <- removeSparseTerms(matrix2, 0.95)
inspect(matrix2)## <<DocumentTermMatrix (documents: 90975, terms: 340)>>
## Non-/sparse entries: 3524175/27407325
## Sparsity : 89%
## Maximal term length: 14
## Weighting : term frequency (tf)
## Sample :
## Terms
## Docs account bank card credit loan mortgage payment payments received told
## 11002 1 5 0 1 68 42 4 4 10 1
## 11343 4 0 1 9 30 6 7 1 2 57
## 25455 79 2 0 11 81 58 29 4 3 0
## 37990 12 6 0 11 41 6 6 4 29 9
## 56802 18 7 0 0 6 20 17 17 21 8
## 60769 0 0 6 5 43 6 7 4 27 18
## 64243 0 0 0 0 26 15 1 10 4 2
## 68621 9 0 0 10 56 0 16 9 10 8
## 70174 8 1 0 7 41 5 59 32 5 9
## 77518 0 0 0 0 837 0 45 1 0 2Convert matrix to dataframe:
matrix_df <- matrix2 %>%
as.matrix() %>%
as.data.frame() Find the most frequent words in text without “stopwords”, use that words for prediction
freq_terms <- data.frame(word = findFreqTerms(matrix2, lowfreq = 5000))We still see some not useful words in the text; let’s remove them since they will not be useful for prediction:
custom_dict <- data.frame(word = c("almost", "anything", "didnt", "even", "however", "never", "still", "use","yet", "also", "cant", "complaint", "done", "due", "fact", "just", "something", "unable", "want", "went", "can", "get", "much", "since", "including", "another", "good", "via", "ask", "call", "help", "need", "problem", "see", "wanted", "around", "find", "assistance", "dont", "every", "may", "issue", "please", "know", "next", "either", "everything", "people", "without", "today", "upon", "additional", "date", "following", "name", "try", "recently", "always", "well", "must", "sure", "per", "able", "finally", "wasnt", "prior", "within", "right", "left", "person"))
train2 <- train
train2$Consumer.complaint.narrative <- gsub(paste0('\\b', custom_dict$word, '\\b', collapse = '|'), '',
train2$Consumer.complaint.narrative )
head(train2)## Product
## 1 Credit card or prepaid card
## 2 Mortgage
## 3 Student loan
## 4 Credit card or prepaid card
## 5 Credit card or prepaid card
## 6 Credit card or prepaid card
## Consumer.complaint.narrative
## 1 initially writing chase bank late charge unauthorized amount many letters complaints closed original account reissued new account sent letter indicating moved charge fraud immediately credited failed months later statement im getting charged interest card pay charges late pay charge make refuse chase refuses many attempts letters write stating feel last step begin case
## 2 ex husband mobile home home mortgage back horrible divorce husband using drugs failing pay bills led divorce mortgage names mortgage one year got trailer bc pay debt working time given birth first child now age happened years ago \n\nnow years filled irs irs alerted caused adverse affect irs taxes \n received l letter irs stated ditech mortgage sending taxable income cause owe irs irs told contact ditech financial dispute request ditech mortgage rescind order taxable income \n\n contacted ditech spoke told recorded anyways \n told settled amount debt mortgage back ditech ever receive asked rescind taxable income debt years ago taxed taxable income debt years ago refused rescind reverse \nnote suffered credit debt ten years removed debt bc year period debt lawfully removed credit go back years later send irs form tax debt years ago charged credit report back year unfair allowed \n disputing kept putting credit years now negatively affected credit way year period \npoint plans taing taxed yrs ago year period passed now way going send taxable income years ago requesting submitted unfairly affected credit past year period violating rights taed debt year
## 3 student accumulated debt student loans school flat lied assumption less half lot credits transferred mention school promised job placement told making lot gave false information thing school put tons debt lied single mother excited degree graduated realize lied school now shut loans forgiven lied given false information basically pressured enrolling known student loans high stayed college prerequisites feel like fool
## 4 come attention citi group actively attempting interfere rights guaranteed constitution nd amendment manipulating denying service certain entities transactions protected constitution expect citi group financial institution controlling nation s activities unpatriotic unamerican behavior expect citi group tend business business stay business social tyranny
## 5 banks new firearm policies run counter laws regulations passed congress infringe discriminate individuals second amendment rights policies endorsed federal government instead business companies respect constitutional rights second amendment federal government take necessary steps review terminate contract citibank unless rescind guidelines
## 6 walmart store card keep cancelled sons college student card cancelled non made small purchase paid made small purchase thought paid online wlamartcom seemed work receive statment probably mail courieri apartment style mailbox system live nevertheless thought credit sesame showed past reporti town time immediately searched walmarts payment site paid account full noticed account reduced mere short now payment included aprox past fees appears maxed account purchase credit dropped score points excellent fair rating
## Company State ZIP.code Submitted.via
## 1 JPMORGAN CHASE & CO. CT 064XX Web
## 2 Ditech Financial LLC GA None Web
## 3 Navient Solutions, LLC. IN 463XX Web
## 4 CITIBANK, N.A. MI 490XX Web
## 5 CITIBANK, N.A. MI 480XX Web
## 6 SYNCHRONY FINANCIAL FL 331XX WebCreate new term matrix:
matrix3 <- DocumentTermMatrix(train2$Consumer.complaint.narrative)
inspect(matrix3)## <<DocumentTermMatrix (documents: 90975, terms: 73748)>>
## Non-/sparse entries: 6244827/6702979473
## Sparsity : 100%
## Maximal term length: 1963
## Weighting : term frequency (tf)
## Sample :
## Terms
## Docs account bank card credit loan mortgage payment payments received told
## 19650 4 34 0 3 49 10 6 7 12 0
## 25455 79 2 0 11 81 58 29 4 3 0
## 27671 2 1 0 1 4 4 8 5 1 0
## 33614 2 32 2 6 1 12 1 0 3 0
## 58908 23 2 0 23 27 0 22 32 3 0
## 64243 0 0 0 0 26 15 1 10 4 2
## 68621 9 0 0 10 56 0 16 9 10 8
## 73540 8 1 0 6 33 61 7 37 3 0
## 77518 0 0 0 0 837 0 45 1 0 2
## 84551 38 0 0 8 70 60 22 16 0 0Remove infrequent words, to reduce matrix size:
matrix3 <- removeSparseTerms(matrix3, 0.95)
inspect(matrix3)## <<DocumentTermMatrix (documents: 90975, terms: 270)>>
## Non-/sparse entries: 2753565/21809685
## Sparsity : 89%
## Maximal term length: 14
## Weighting : term frequency (tf)
## Sample :
## Terms
## Docs account bank card credit loan mortgage payment payments received told
## 11002 1 5 0 1 68 42 4 4 10 1
## 15879 2 73 0 0 19 12 2 6 38 8
## 19650 4 34 0 3 49 10 6 7 12 0
## 25455 79 2 0 11 81 58 29 4 3 0
## 37990 12 6 0 11 41 6 6 4 29 9
## 64243 0 0 0 0 26 15 1 10 4 2
## 68621 9 0 0 10 56 0 16 9 10 8
## 70174 8 1 0 7 41 5 59 32 5 9
## 77518 0 0 0 0 837 0 45 1 0 2
## 84551 38 0 0 8 70 60 22 16 0 0Convert matrix to dataframe:
matrix_df <- matrix3 %>%
as.matrix() %>%
as.data.frame() Find the most frequent words in text without “stopwords”:
freq_terms <- data.frame(word = findFreqTerms(matrix3, lowfreq = 5000))Merge with train2 dataset, to get the Product column back:
matrix_df <- matrix_df %>%
as.matrix() %>%
as.data.frame() %>%
bind_cols(Product=train2$Product) %>%
select(Product, everything())Get a snaphost of the matrix, and some information:
skim(matrix_df)| Name | matrix_df |
| Number of rows | 90975 |
| Number of columns | 271 |
| _______________________ | |
| Column type frequency: | |
| factor | 1 |
| numeric | 270 |
| ________________________ | |
| Group variables | None |
Variable type: factor
| skim_variable | n_missing | complete_rate | ordered | n_unique | top_counts |
|---|---|---|---|---|---|
| Product | 0 | 1 | FALSE | 4 | Cre: 38294, Mor: 30957, Stu: 12485, Veh: 9239 |
Variable type: numeric
| skim_variable | n_missing | complete_rate | mean | sd | p0 | p25 | p50 | p75 | p100 | hist |
|---|---|---|---|---|---|---|---|---|---|---|
| account | 0 | 1 | 1.30 | 2.56 | 0 | 0 | 0 | 2 | 165 | ▇▁▁▁▁ |
| amount | 0 | 1 | 0.49 | 1.30 | 0 | 0 | 0 | 1 | 98 | ▇▁▁▁▁ |
| bank | 0 | 1 | 0.69 | 2.02 | 0 | 0 | 0 | 1 | 82 | ▇▁▁▁▁ |
| card | 0 | 1 | 1.03 | 2.42 | 0 | 0 | 0 | 1 | 77 | ▇▁▁▁▁ |
| case | 0 | 1 | 0.15 | 0.67 | 0 | 0 | 0 | 0 | 36 | ▇▁▁▁▁ |
| charge | 0 | 1 | 0.25 | 0.96 | 0 | 0 | 0 | 0 | 61 | ▇▁▁▁▁ |
| charged | 0 | 1 | 0.18 | 0.62 | 0 | 0 | 0 | 0 | 30 | ▇▁▁▁▁ |
| charges | 0 | 1 | 0.24 | 1.05 | 0 | 0 | 0 | 0 | 51 | ▇▁▁▁▁ |
| chase | 0 | 1 | 0.24 | 1.56 | 0 | 0 | 0 | 0 | 59 | ▇▁▁▁▁ |
| closed | 0 | 1 | 0.21 | 0.71 | 0 | 0 | 0 | 0 | 25 | ▇▁▁▁▁ |
| feel | 0 | 1 | 0.10 | 0.37 | 0 | 0 | 0 | 0 | 15 | ▇▁▁▁▁ |
| fraud | 0 | 1 | 0.18 | 0.91 | 0 | 0 | 0 | 0 | 38 | ▇▁▁▁▁ |
| getting | 0 | 1 | 0.11 | 0.38 | 0 | 0 | 0 | 0 | 11 | ▇▁▁▁▁ |
| immediately | 0 | 1 | 0.12 | 0.42 | 0 | 0 | 0 | 0 | 16 | ▇▁▁▁▁ |
| interest | 0 | 1 | 0.46 | 1.44 | 0 | 0 | 0 | 0 | 70 | ▇▁▁▁▁ |
| last | 0 | 1 | 0.19 | 0.53 | 0 | 0 | 0 | 0 | 14 | ▇▁▁▁▁ |
| late | 0 | 1 | 0.43 | 1.20 | 0 | 0 | 0 | 0 | 37 | ▇▁▁▁▁ |
| later | 0 | 1 | 0.15 | 0.46 | 0 | 0 | 0 | 0 | 12 | ▇▁▁▁▁ |
| letter | 0 | 1 | 0.44 | 1.25 | 0 | 0 | 0 | 0 | 44 | ▇▁▁▁▁ |
| make | 0 | 1 | 0.37 | 0.90 | 0 | 0 | 0 | 0 | 24 | ▇▁▁▁▁ |
| many | 0 | 1 | 0.12 | 0.45 | 0 | 0 | 0 | 0 | 12 | ▇▁▁▁▁ |
| months | 0 | 1 | 0.33 | 0.82 | 0 | 0 | 0 | 0 | 19 | ▇▁▁▁▁ |
| new | 0 | 1 | 0.32 | 0.94 | 0 | 0 | 0 | 0 | 32 | ▇▁▁▁▁ |
| original | 0 | 1 | 0.11 | 0.49 | 0 | 0 | 0 | 0 | 22 | ▇▁▁▁▁ |
| pay | 0 | 1 | 0.57 | 1.20 | 0 | 0 | 0 | 1 | 28 | ▇▁▁▁▁ |
| sent | 0 | 1 | 0.47 | 1.11 | 0 | 0 | 0 | 1 | 53 | ▇▁▁▁▁ |
| statement | 0 | 1 | 0.26 | 0.93 | 0 | 0 | 0 | 0 | 30 | ▇▁▁▁▁ |
| stating | 0 | 1 | 0.16 | 0.53 | 0 | 0 | 0 | 0 | 27 | ▇▁▁▁▁ |
| ago | 0 | 1 | 0.08 | 0.33 | 0 | 0 | 0 | 0 | 12 | ▇▁▁▁▁ |
| asked | 0 | 1 | 0.36 | 0.93 | 0 | 0 | 0 | 0 | 30 | ▇▁▁▁▁ |
| back | 0 | 1 | 0.56 | 1.23 | 0 | 0 | 0 | 1 | 28 | ▇▁▁▁▁ |
| contact | 0 | 1 | 0.20 | 0.88 | 0 | 0 | 0 | 0 | 187 | ▇▁▁▁▁ |
| contacted | 0 | 1 | 0.24 | 0.65 | 0 | 0 | 0 | 0 | 23 | ▇▁▁▁▁ |
| credit | 0 | 1 | 1.36 | 2.50 | 0 | 0 | 0 | 2 | 100 | ▇▁▁▁▁ |
| debt | 0 | 1 | 0.14 | 0.80 | 0 | 0 | 0 | 0 | 48 | ▇▁▁▁▁ |
| dispute | 0 | 1 | 0.20 | 1.04 | 0 | 0 | 0 | 0 | 75 | ▇▁▁▁▁ |
| ever | 0 | 1 | 0.08 | 0.33 | 0 | 0 | 0 | 0 | 18 | ▇▁▁▁▁ |
| financial | 0 | 1 | 0.19 | 0.78 | 0 | 0 | 0 | 0 | 31 | ▇▁▁▁▁ |
| first | 0 | 1 | 0.25 | 0.70 | 0 | 0 | 0 | 0 | 21 | ▇▁▁▁▁ |
| form | 0 | 1 | 0.08 | 0.48 | 0 | 0 | 0 | 0 | 32 | ▇▁▁▁▁ |
| given | 0 | 1 | 0.12 | 0.44 | 0 | 0 | 0 | 0 | 21 | ▇▁▁▁▁ |
| going | 0 | 1 | 0.19 | 0.60 | 0 | 0 | 0 | 0 | 42 | ▇▁▁▁▁ |
| got | 0 | 1 | 0.19 | 0.60 | 0 | 0 | 0 | 0 | 21 | ▇▁▁▁▁ |
| happened | 0 | 1 | 0.08 | 0.33 | 0 | 0 | 0 | 0 | 10 | ▇▁▁▁▁ |
| home | 0 | 1 | 0.40 | 1.36 | 0 | 0 | 0 | 0 | 48 | ▇▁▁▁▁ |
| income | 0 | 1 | 0.13 | 0.74 | 0 | 0 | 0 | 0 | 56 | ▇▁▁▁▁ |
| mortgage | 0 | 1 | 0.81 | 2.36 | 0 | 0 | 0 | 1 | 130 | ▇▁▁▁▁ |
| note | 0 | 1 | 0.09 | 0.70 | 0 | 0 | 0 | 0 | 64 | ▇▁▁▁▁ |
| now | 0 | 1 | 0.44 | 0.88 | 0 | 0 | 0 | 1 | 21 | ▇▁▁▁▁ |
| one | 0 | 1 | 0.58 | 1.42 | 0 | 0 | 0 | 1 | 91 | ▇▁▁▁▁ |
| order | 0 | 1 | 0.12 | 0.59 | 0 | 0 | 0 | 0 | 57 | ▇▁▁▁▁ |
| owe | 0 | 1 | 0.08 | 0.37 | 0 | 0 | 0 | 0 | 12 | ▇▁▁▁▁ |
| past | 0 | 1 | 0.14 | 0.54 | 0 | 0 | 0 | 0 | 33 | ▇▁▁▁▁ |
| period | 0 | 1 | 0.09 | 0.47 | 0 | 0 | 0 | 0 | 34 | ▇▁▁▁▁ |
| point | 0 | 1 | 0.11 | 0.48 | 0 | 0 | 0 | 0 | 21 | ▇▁▁▁▁ |
| receive | 0 | 1 | 0.19 | 0.57 | 0 | 0 | 0 | 0 | 20 | ▇▁▁▁▁ |
| received | 0 | 1 | 0.74 | 1.45 | 0 | 0 | 0 | 1 | 49 | ▇▁▁▁▁ |
| refused | 0 | 1 | 0.11 | 0.42 | 0 | 0 | 0 | 0 | 11 | ▇▁▁▁▁ |
| removed | 0 | 1 | 0.08 | 0.39 | 0 | 0 | 0 | 0 | 13 | ▇▁▁▁▁ |
| report | 0 | 1 | 0.25 | 0.76 | 0 | 0 | 0 | 0 | 34 | ▇▁▁▁▁ |
| request | 0 | 1 | 0.25 | 0.88 | 0 | 0 | 0 | 0 | 48 | ▇▁▁▁▁ |
| requesting | 0 | 1 | 0.07 | 0.31 | 0 | 0 | 0 | 0 | 9 | ▇▁▁▁▁ |
| send | 0 | 1 | 0.19 | 0.60 | 0 | 0 | 0 | 0 | 17 | ▇▁▁▁▁ |
| spoke | 0 | 1 | 0.21 | 0.68 | 0 | 0 | 0 | 0 | 38 | ▇▁▁▁▁ |
| stated | 0 | 1 | 0.24 | 0.86 | 0 | 0 | 0 | 0 | 52 | ▇▁▁▁▁ |
| submitted | 0 | 1 | 0.12 | 0.53 | 0 | 0 | 0 | 0 | 27 | ▇▁▁▁▁ |
| time | 0 | 1 | 0.69 | 1.25 | 0 | 0 | 0 | 1 | 50 | ▇▁▁▁▁ |
| told | 0 | 1 | 0.84 | 1.70 | 0 | 0 | 0 | 1 | 57 | ▇▁▁▁▁ |
| using | 0 | 1 | 0.09 | 0.36 | 0 | 0 | 0 | 0 | 10 | ▇▁▁▁▁ |
| way | 0 | 1 | 0.14 | 0.45 | 0 | 0 | 0 | 0 | 10 | ▇▁▁▁▁ |
| working | 0 | 1 | 0.08 | 0.35 | 0 | 0 | 0 | 0 | 14 | ▇▁▁▁▁ |
| year | 0 | 1 | 0.21 | 0.65 | 0 | 0 | 0 | 0 | 17 | ▇▁▁▁▁ |
| years | 0 | 1 | 0.29 | 0.82 | 0 | 0 | 0 | 0 | 30 | ▇▁▁▁▁ |
| gave | 0 | 1 | 0.09 | 0.37 | 0 | 0 | 0 | 0 | 10 | ▇▁▁▁▁ |
| information | 0 | 1 | 0.47 | 1.24 | 0 | 0 | 0 | 0 | 60 | ▇▁▁▁▁ |
| like | 0 | 1 | 0.21 | 0.60 | 0 | 0 | 0 | 0 | 23 | ▇▁▁▁▁ |
| loans | 0 | 1 | 0.28 | 1.22 | 0 | 0 | 0 | 0 | 58 | ▇▁▁▁▁ |
| making | 0 | 1 | 0.14 | 0.45 | 0 | 0 | 0 | 0 | 14 | ▇▁▁▁▁ |
| put | 0 | 1 | 0.16 | 0.50 | 0 | 0 | 0 | 0 | 10 | ▇▁▁▁▁ |
| student | 0 | 1 | 0.14 | 0.72 | 0 | 0 | 0 | 0 | 53 | ▇▁▁▁▁ |
| transferred | 0 | 1 | 0.11 | 0.45 | 0 | 0 | 0 | 0 | 12 | ▇▁▁▁▁ |
| business | 0 | 1 | 0.17 | 0.65 | 0 | 0 | 0 | 0 | 36 | ▇▁▁▁▁ |
| service | 0 | 1 | 0.31 | 0.88 | 0 | 0 | 0 | 0 | 47 | ▇▁▁▁▁ |
| federal | 0 | 1 | 0.10 | 0.65 | 0 | 0 | 0 | 0 | 52 | ▇▁▁▁▁ |
| instead | 0 | 1 | 0.08 | 0.34 | 0 | 0 | 0 | 0 | 7 | ▇▁▁▁▁ |
| review | 0 | 1 | 0.11 | 0.54 | 0 | 0 | 0 | 0 | 28 | ▇▁▁▁▁ |
| second | 0 | 1 | 0.09 | 0.42 | 0 | 0 | 0 | 0 | 17 | ▇▁▁▁▁ |
| take | 0 | 1 | 0.20 | 0.56 | 0 | 0 | 0 | 0 | 16 | ▇▁▁▁▁ |
| fees | 0 | 1 | 0.24 | 0.98 | 0 | 0 | 0 | 0 | 67 | ▇▁▁▁▁ |
| full | 0 | 1 | 0.18 | 0.62 | 0 | 0 | 0 | 0 | 25 | ▇▁▁▁▁ |
| keep | 0 | 1 | 0.11 | 0.40 | 0 | 0 | 0 | 0 | 13 | ▇▁▁▁▁ |
| made | 0 | 1 | 0.53 | 1.10 | 0 | 0 | 0 | 1 | 31 | ▇▁▁▁▁ |
| 0 | 1 | 0.15 | 0.54 | 0 | 0 | 0 | 0 | 16 | ▇▁▁▁▁ | |
| online | 0 | 1 | 0.18 | 0.64 | 0 | 0 | 0 | 0 | 25 | ▇▁▁▁▁ |
| paid | 0 | 1 | 0.49 | 1.12 | 0 | 0 | 0 | 1 | 30 | ▇▁▁▁▁ |
| payment | 0 | 1 | 1.38 | 2.84 | 0 | 0 | 0 | 2 | 81 | ▇▁▁▁▁ |
| purchase | 0 | 1 | 0.15 | 0.63 | 0 | 0 | 0 | 0 | 20 | ▇▁▁▁▁ |
| score | 0 | 1 | 0.11 | 0.49 | 0 | 0 | 0 | 0 | 20 | ▇▁▁▁▁ |
| system | 0 | 1 | 0.10 | 0.48 | 0 | 0 | 0 | 0 | 14 | ▇▁▁▁▁ |
| thought | 0 | 1 | 0.06 | 0.27 | 0 | 0 | 0 | 0 | 6 | ▇▁▁▁▁ |
| work | 0 | 1 | 0.18 | 0.62 | 0 | 0 | 0 | 0 | 27 | ▇▁▁▁▁ |
| answer | 0 | 1 | 0.08 | 0.36 | 0 | 0 | 0 | 0 | 14 | ▇▁▁▁▁ |
| applied | 0 | 1 | 0.20 | 0.69 | 0 | 0 | 0 | 0 | 40 | ▇▁▁▁▁ |
| called | 0 | 1 | 0.68 | 1.28 | 0 | 0 | 0 | 1 | 32 | ▇▁▁▁▁ |
| check | 0 | 1 | 0.28 | 1.16 | 0 | 0 | 0 | 0 | 38 | ▇▁▁▁▁ |
| fee | 0 | 1 | 0.23 | 0.95 | 0 | 0 | 0 | 0 | 38 | ▇▁▁▁▁ |
| previous | 0 | 1 | 0.08 | 0.35 | 0 | 0 | 0 | 0 | 12 | ▇▁▁▁▁ |
| provided | 0 | 1 | 0.17 | 0.62 | 0 | 0 | 0 | 0 | 31 | ▇▁▁▁▁ |
| representative | 0 | 1 | 0.20 | 0.76 | 0 | 0 | 0 | 0 | 24 | ▇▁▁▁▁ |
| response | 0 | 1 | 0.15 | 0.63 | 0 | 0 | 0 | 0 | 50 | ▇▁▁▁▁ |
| started | 0 | 1 | 0.09 | 0.37 | 0 | 0 | 0 | 0 | 22 | ▇▁▁▁▁ |
| three | 0 | 1 | 0.10 | 0.43 | 0 | 0 | 0 | 0 | 28 | ▇▁▁▁▁ |
| calls | 0 | 1 | 0.15 | 0.53 | 0 | 0 | 0 | 0 | 24 | ▇▁▁▁▁ |
| company | 0 | 1 | 0.57 | 1.29 | 0 | 0 | 0 | 1 | 50 | ▇▁▁▁▁ |
| days | 0 | 1 | 0.38 | 0.93 | 0 | 0 | 0 | 0 | 30 | ▇▁▁▁▁ |
| end | 0 | 1 | 0.12 | 0.46 | 0 | 0 | 0 | 0 | 15 | ▇▁▁▁▁ |
| hours | 0 | 1 | 0.09 | 0.40 | 0 | 0 | 0 | 0 | 18 | ▇▁▁▁▁ |
| payments | 0 | 1 | 0.81 | 1.80 | 0 | 0 | 0 | 1 | 45 | ▇▁▁▁▁ |
| phone | 0 | 1 | 0.36 | 0.93 | 0 | 0 | 0 | 0 | 34 | ▇▁▁▁▁ |
| agreement | 0 | 1 | 0.10 | 0.63 | 0 | 0 | 0 | 0 | 32 | ▇▁▁▁▁ |
| asking | 0 | 1 | 0.09 | 0.36 | 0 | 0 | 0 | 0 | 10 | ▇▁▁▁▁ |
| loan | 0 | 1 | 1.29 | 4.07 | 0 | 0 | 0 | 1 | 837 | ▇▁▁▁▁ |
| office | 0 | 1 | 0.10 | 0.52 | 0 | 0 | 0 | 0 | 22 | ▇▁▁▁▁ |
| paperwork | 0 | 1 | 0.09 | 0.51 | 0 | 0 | 0 | 0 | 17 | ▇▁▁▁▁ |
| paying | 0 | 1 | 0.17 | 0.53 | 0 | 0 | 0 | 0 | 20 | ▇▁▁▁▁ |
| regarding | 0 | 1 | 0.12 | 0.46 | 0 | 0 | 0 | 0 | 14 | ▇▁▁▁▁ |
| statements | 0 | 1 | 0.11 | 0.54 | 0 | 0 | 0 | 0 | 27 | ▇▁▁▁▁ |
| understand | 0 | 1 | 0.08 | 0.32 | 0 | 0 | 0 | 0 | 9 | ▇▁▁▁▁ |
| advised | 0 | 1 | 0.12 | 0.68 | 0 | 0 | 0 | 0 | 37 | ▇▁▁▁▁ |
| balance | 0 | 1 | 0.42 | 1.32 | 0 | 0 | 0 | 0 | 81 | ▇▁▁▁▁ |
| money | 0 | 1 | 0.34 | 1.00 | 0 | 0 | 0 | 0 | 44 | ▇▁▁▁▁ |
| monthly | 0 | 1 | 0.21 | 0.76 | 0 | 0 | 0 | 0 | 30 | ▇▁▁▁▁ |
| rep | 0 | 1 | 0.09 | 0.52 | 0 | 0 | 0 | 0 | 24 | ▇▁▁▁▁ |
| reporting | 0 | 1 | 0.11 | 0.49 | 0 | 0 | 0 | 0 | 23 | ▇▁▁▁▁ |
| returned | 0 | 1 | 0.10 | 0.48 | 0 | 0 | 0 | 0 | 39 | ▇▁▁▁▁ |
| website | 0 | 1 | 0.11 | 0.48 | 0 | 0 | 0 | 0 | 13 | ▇▁▁▁▁ |
| week | 0 | 1 | 0.10 | 0.39 | 0 | 0 | 0 | 0 | 21 | ▇▁▁▁▁ |
| complete | 0 | 1 | 0.07 | 0.36 | 0 | 0 | 0 | 0 | 13 | ▇▁▁▁▁ |
| department | 0 | 1 | 0.19 | 0.76 | 0 | 0 | 0 | 0 | 32 | ▇▁▁▁▁ |
| documents | 0 | 1 | 0.21 | 0.95 | 0 | 0 | 0 | 0 | 68 | ▇▁▁▁▁ |
| modification | 0 | 1 | 0.20 | 1.18 | 0 | 0 | 0 | 0 | 42 | ▇▁▁▁▁ |
| month | 0 | 1 | 0.37 | 0.92 | 0 | 0 | 0 | 0 | 21 | ▇▁▁▁▁ |
| number | 0 | 1 | 0.30 | 0.95 | 0 | 0 | 0 | 0 | 26 | ▇▁▁▁▁ |
| process | 0 | 1 | 0.22 | 0.72 | 0 | 0 | 0 | 0 | 21 | ▇▁▁▁▁ |
| stop | 0 | 1 | 0.07 | 0.34 | 0 | 0 | 0 | 0 | 14 | ▇▁▁▁▁ |
| funds | 0 | 1 | 0.14 | 0.71 | 0 | 0 | 0 | 0 | 45 | ▇▁▁▁▁ |
| give | 0 | 1 | 0.13 | 0.43 | 0 | 0 | 0 | 0 | 10 | ▇▁▁▁▁ |
| matter | 0 | 1 | 0.11 | 0.43 | 0 | 0 | 0 | 0 | 16 | ▇▁▁▁▁ |
| provide | 0 | 1 | 0.16 | 0.58 | 0 | 0 | 0 | 0 | 22 | ▇▁▁▁▁ |
| say | 0 | 1 | 0.09 | 0.35 | 0 | 0 | 0 | 0 | 10 | ▇▁▁▁▁ |
| thank | 0 | 1 | 0.08 | 0.36 | 0 | 0 | 0 | 0 | 25 | ▇▁▁▁▁ |
| times | 0 | 1 | 0.24 | 0.58 | 0 | 0 | 0 | 0 | 16 | ▇▁▁▁▁ |
| transaction | 0 | 1 | 0.10 | 0.62 | 0 | 0 | 0 | 0 | 36 | ▇▁▁▁▁ |
| believe | 0 | 1 | 0.12 | 0.43 | 0 | 0 | 0 | 0 | 11 | ▇▁▁▁▁ |
| currently | 0 | 1 | 0.06 | 0.28 | 0 | 0 | 0 | 0 | 10 | ▇▁▁▁▁ |
| program | 0 | 1 | 0.11 | 0.67 | 0 | 0 | 0 | 0 | 33 | ▇▁▁▁▁ |
| services | 0 | 1 | 0.10 | 0.60 | 0 | 0 | 0 | 0 | 48 | ▇▁▁▁▁ |
| several | 0 | 1 | 0.19 | 0.53 | 0 | 0 | 0 | 0 | 11 | ▇▁▁▁▁ |
| address | 0 | 1 | 0.16 | 0.75 | 0 | 0 | 0 | 0 | 68 | ▇▁▁▁▁ |
| calling | 0 | 1 | 0.07 | 0.31 | 0 | 0 | 0 | 0 | 11 | ▇▁▁▁▁ |
| claim | 0 | 1 | 0.13 | 0.68 | 0 | 0 | 0 | 0 | 44 | ▇▁▁▁▁ |
| filed | 0 | 1 | 0.12 | 0.47 | 0 | 0 | 0 | 0 | 21 | ▇▁▁▁▁ |
| informed | 0 | 1 | 0.17 | 0.66 | 0 | 0 | 0 | 0 | 22 | ▇▁▁▁▁ |
| let | 0 | 1 | 0.07 | 0.34 | 0 | 0 | 0 | 0 | 21 | ▇▁▁▁▁ |
| someone | 0 | 1 | 0.13 | 0.45 | 0 | 0 | 0 | 0 | 10 | ▇▁▁▁▁ |
| status | 0 | 1 | 0.08 | 0.60 | 0 | 0 | 0 | 0 | 130 | ▇▁▁▁▁ |
| used | 0 | 1 | 0.14 | 0.49 | 0 | 0 | 0 | 0 | 13 | ▇▁▁▁▁ |
| continue | 0 | 1 | 0.08 | 0.34 | 0 | 0 | 0 | 0 | 10 | ▇▁▁▁▁ |
| found | 0 | 1 | 0.11 | 0.39 | 0 | 0 | 0 | 0 | 9 | ▇▁▁▁▁ |
| taken | 0 | 1 | 0.09 | 0.37 | 0 | 0 | 0 | 0 | 14 | ▇▁▁▁▁ |
| based | 0 | 1 | 0.09 | 0.40 | 0 | 0 | 0 | 0 | 12 | ▇▁▁▁▁ |
| cfpb | 0 | 1 | 0.09 | 0.58 | 0 | 0 | 0 | 0 | 43 | ▇▁▁▁▁ |
| rate | 0 | 1 | 0.19 | 0.94 | 0 | 0 | 0 | 0 | 62 | ▇▁▁▁▁ |
| refund | 0 | 1 | 0.15 | 0.74 | 0 | 0 | 0 | 0 | 32 | ▇▁▁▁▁ |
| requested | 0 | 1 | 0.20 | 0.63 | 0 | 0 | 0 | 0 | 40 | ▇▁▁▁▁ |
| resolve | 0 | 1 | 0.09 | 0.36 | 0 | 0 | 0 | 0 | 14 | ▇▁▁▁▁ |
| total | 0 | 1 | 0.11 | 0.55 | 0 | 0 | 0 | 0 | 65 | ▇▁▁▁▁ |
| tried | 0 | 1 | 0.16 | 0.47 | 0 | 0 | 0 | 0 | 11 | ▇▁▁▁▁ |
| change | 0 | 1 | 0.09 | 0.44 | 0 | 0 | 0 | 0 | 16 | ▇▁▁▁▁ |
| changed | 0 | 1 | 0.07 | 0.34 | 0 | 0 | 0 | 0 | 12 | ▇▁▁▁▁ |
| day | 0 | 1 | 0.24 | 0.69 | 0 | 0 | 0 | 0 | 15 | ▇▁▁▁▁ |
| notified | 0 | 1 | 0.06 | 0.31 | 0 | 0 | 0 | 0 | 13 | ▇▁▁▁▁ |
| wrong | 0 | 1 | 0.08 | 0.35 | 0 | 0 | 0 | 0 | 8 | ▇▁▁▁▁ |
| close | 0 | 1 | 0.12 | 0.55 | 0 | 0 | 0 | 0 | 32 | ▇▁▁▁▁ |
| offered | 0 | 1 | 0.07 | 0.33 | 0 | 0 | 0 | 0 | 17 | ▇▁▁▁▁ |
| opened | 0 | 1 | 0.08 | 0.39 | 0 | 0 | 0 | 0 | 13 | ▇▁▁▁▁ |
| bill | 0 | 1 | 0.15 | 0.66 | 0 | 0 | 0 | 0 | 38 | ▇▁▁▁▁ |
| car | 0 | 1 | 0.20 | 1.23 | 0 | 0 | 0 | 0 | 80 | ▇▁▁▁▁ |
| hold | 0 | 1 | 0.11 | 0.46 | 0 | 0 | 0 | 0 | 19 | ▇▁▁▁▁ |
| reason | 0 | 1 | 0.11 | 0.41 | 0 | 0 | 0 | 0 | 11 | ▇▁▁▁▁ |
| said | 0 | 1 | 0.50 | 1.36 | 0 | 0 | 0 | 0 | 46 | ▇▁▁▁▁ |
| agent | 0 | 1 | 0.10 | 0.61 | 0 | 0 | 0 | 0 | 31 | ▇▁▁▁▁ |
| fraudulent | 0 | 1 | 0.11 | 0.59 | 0 | 0 | 0 | 0 | 19 | ▇▁▁▁▁ |
| insurance | 0 | 1 | 0.24 | 1.26 | 0 | 0 | 0 | 0 | 57 | ▇▁▁▁▁ |
| property | 0 | 1 | 0.21 | 1.16 | 0 | 0 | 0 | 0 | 64 | ▇▁▁▁▁ |
| servicing | 0 | 1 | 0.13 | 0.79 | 0 | 0 | 0 | 0 | 65 | ▇▁▁▁▁ |
| already | 0 | 1 | 0.13 | 0.43 | 0 | 0 | 0 | 0 | 11 | ▇▁▁▁▁ |
| correct | 0 | 1 | 0.11 | 0.44 | 0 | 0 | 0 | 0 | 15 | ▇▁▁▁▁ |
| current | 0 | 1 | 0.14 | 0.54 | 0 | 0 | 0 | 0 | 20 | ▇▁▁▁▁ |
| dated | 0 | 1 | 0.08 | 0.46 | 0 | 0 | 0 | 0 | 26 | ▇▁▁▁▁ |
| foreclosure | 0 | 1 | 0.14 | 0.77 | 0 | 0 | 0 | 0 | 39 | ▇▁▁▁▁ |
| law | 0 | 1 | 0.08 | 0.62 | 0 | 0 | 0 | 0 | 79 | ▇▁▁▁▁ |
| part | 0 | 1 | 0.07 | 0.33 | 0 | 0 | 0 | 0 | 15 | ▇▁▁▁▁ |
| place | 0 | 1 | 0.07 | 0.32 | 0 | 0 | 0 | 0 | 16 | ▇▁▁▁▁ |
| required | 0 | 1 | 0.10 | 0.43 | 0 | 0 | 0 | 0 | 21 | ▇▁▁▁▁ |
| came | 0 | 1 | 0.07 | 0.32 | 0 | 0 | 0 | 0 | 9 | ▇▁▁▁▁ |
| house | 0 | 1 | 0.13 | 0.68 | 0 | 0 | 0 | 0 | 36 | ▇▁▁▁▁ |
| lender | 0 | 1 | 0.11 | 0.66 | 0 | 0 | 0 | 0 | 29 | ▇▁▁▁▁ |
| application | 0 | 1 | 0.17 | 0.85 | 0 | 0 | 0 | 0 | 32 | ▇▁▁▁▁ |
| apply | 0 | 1 | 0.08 | 0.38 | 0 | 0 | 0 | 0 | 16 | ▇▁▁▁▁ |
| needed | 0 | 1 | 0.15 | 0.54 | 0 | 0 | 0 | 0 | 18 | ▇▁▁▁▁ |
| plan | 0 | 1 | 0.12 | 0.72 | 0 | 0 | 0 | 0 | 77 | ▇▁▁▁▁ |
| situation | 0 | 1 | 0.12 | 0.45 | 0 | 0 | 0 | 0 | 21 | ▇▁▁▁▁ |
| customer | 0 | 1 | 0.30 | 0.88 | 0 | 0 | 0 | 0 | 28 | ▇▁▁▁▁ |
| long | 0 | 1 | 0.07 | 0.30 | 0 | 0 | 0 | 0 | 7 | ▇▁▁▁▁ |
| multiple | 0 | 1 | 0.10 | 0.40 | 0 | 0 | 0 | 0 | 11 | ▇▁▁▁▁ |
| saying | 0 | 1 | 0.11 | 0.39 | 0 | 0 | 0 | 0 | 9 | ▇▁▁▁▁ |
| terms | 0 | 1 | 0.09 | 0.49 | 0 | 0 | 0 | 0 | 20 | ▇▁▁▁▁ |
| weeks | 0 | 1 | 0.11 | 0.41 | 0 | 0 | 0 | 0 | 13 | ▇▁▁▁▁ |
| attached | 0 | 1 | 0.11 | 0.51 | 0 | 0 | 0 | 0 | 21 | ▇▁▁▁▁ |
| available | 0 | 1 | 0.08 | 0.39 | 0 | 0 | 0 | 0 | 23 | ▇▁▁▁▁ |
| denied | 0 | 1 | 0.11 | 0.46 | 0 | 0 | 0 | 0 | 16 | ▇▁▁▁▁ |
| 0 | 1 | 0.30 | 1.07 | 0 | 0 | 0 | 0 | 54 | ▇▁▁▁▁ | |
| error | 0 | 1 | 0.10 | 0.51 | 0 | 0 | 0 | 0 | 29 | ▇▁▁▁▁ |
| set | 0 | 1 | 0.09 | 0.43 | 0 | 0 | 0 | 0 | 14 | ▇▁▁▁▁ |
| closing | 0 | 1 | 0.14 | 0.88 | 0 | 0 | 0 | 0 | 51 | ▇▁▁▁▁ |
| sold | 0 | 1 | 0.08 | 0.39 | 0 | 0 | 0 | 0 | 15 | ▇▁▁▁▁ |
| tell | 0 | 1 | 0.09 | 0.39 | 0 | 0 | 0 | 0 | 19 | ▇▁▁▁▁ |
| escrow | 0 | 1 | 0.22 | 1.20 | 0 | 0 | 0 | 0 | 55 | ▇▁▁▁▁ |
| took | 0 | 1 | 0.14 | 0.45 | 0 | 0 | 0 | 0 | 11 | ▇▁▁▁▁ |
| attorney | 0 | 1 | 0.10 | 0.62 | 0 | 0 | 0 | 0 | 41 | ▇▁▁▁▁ |
| ive | 0 | 1 | 0.12 | 0.50 | 0 | 0 | 0 | 0 | 16 | ▇▁▁▁▁ |
| resolved | 0 | 1 | 0.07 | 0.34 | 0 | 0 | 0 | 0 | 17 | ▇▁▁▁▁ |
| reported | 0 | 1 | 0.11 | 0.44 | 0 | 0 | 0 | 0 | 22 | ▇▁▁▁▁ |
| state | 0 | 1 | 0.09 | 0.51 | 0 | 0 | 0 | 0 | 49 | ▇▁▁▁▁ |
| states | 0 | 1 | 0.08 | 0.52 | 0 | 0 | 0 | 0 | 59 | ▇▁▁▁▁ |
| file | 0 | 1 | 0.13 | 0.56 | 0 | 0 | 0 | 0 | 35 | ▇▁▁▁▁ |
| though | 0 | 1 | 0.10 | 0.38 | 0 | 0 | 0 | 0 | 10 | ▇▁▁▁▁ |
| two | 0 | 1 | 0.23 | 0.66 | 0 | 0 | 0 | 0 | 32 | ▇▁▁▁▁ |
| different | 0 | 1 | 0.12 | 0.43 | 0 | 0 | 0 | 0 | 14 | ▇▁▁▁▁ |
| longer | 0 | 1 | 0.07 | 0.31 | 0 | 0 | 0 | 0 | 10 | ▇▁▁▁▁ |
| receiving | 0 | 1 | 0.07 | 0.31 | 0 | 0 | 0 | 0 | 9 | ▇▁▁▁▁ |
| accounts | 0 | 1 | 0.12 | 0.61 | 0 | 0 | 0 | 0 | 27 | ▇▁▁▁▁ |
| consumer | 0 | 1 | 0.11 | 0.61 | 0 | 0 | 0 | 0 | 51 | ▇▁▁▁▁ |
| documentation | 0 | 1 | 0.11 | 0.51 | 0 | 0 | 0 | 0 | 22 | ▇▁▁▁▁ |
| nothing | 0 | 1 | 0.15 | 0.46 | 0 | 0 | 0 | 0 | 11 | ▇▁▁▁▁ |
| return | 0 | 1 | 0.09 | 0.46 | 0 | 0 | 0 | 0 | 21 | ▇▁▁▁▁ |
| cards | 0 | 1 | 0.12 | 0.62 | 0 | 0 | 0 | 0 | 35 | ▇▁▁▁▁ |
| line | 0 | 1 | 0.10 | 0.47 | 0 | 0 | 0 | 0 | 25 | ▇▁▁▁▁ |
| notice | 0 | 1 | 0.13 | 0.58 | 0 | 0 | 0 | 0 | 29 | ▇▁▁▁▁ |
| signed | 0 | 1 | 0.11 | 0.47 | 0 | 0 | 0 | 0 | 45 | ▇▁▁▁▁ |
| explained | 0 | 1 | 0.11 | 0.49 | 0 | 0 | 0 | 0 | 16 | ▇▁▁▁▁ |
| proof | 0 | 1 | 0.11 | 0.51 | 0 | 0 | 0 | 0 | 28 | ▇▁▁▁▁ |
| approved | 0 | 1 | 0.10 | 0.46 | 0 | 0 | 0 | 0 | 15 | ▇▁▁▁▁ |
| offer | 0 | 1 | 0.12 | 0.67 | 0 | 0 | 0 | 0 | 37 | ▇▁▁▁▁ |
| supervisor | 0 | 1 | 0.13 | 0.59 | 0 | 0 | 0 | 0 | 27 | ▇▁▁▁▁ |
| trying | 0 | 1 | 0.15 | 0.46 | 0 | 0 | 0 | 0 | 17 | ▇▁▁▁▁ |
| agreed | 0 | 1 | 0.08 | 0.38 | 0 | 0 | 0 | 0 | 12 | ▇▁▁▁▁ |
| copy | 0 | 1 | 0.10 | 0.51 | 0 | 0 | 0 | 0 | 45 | ▇▁▁▁▁ |
| issues | 0 | 1 | 0.09 | 0.40 | 0 | 0 | 0 | 0 | 19 | ▇▁▁▁▁ |
| history | 0 | 1 | 0.08 | 0.43 | 0 | 0 | 0 | 0 | 31 | ▇▁▁▁▁ |
| show | 0 | 1 | 0.07 | 0.33 | 0 | 0 | 0 | 0 | 19 | ▇▁▁▁▁ |
| transfer | 0 | 1 | 0.11 | 0.66 | 0 | 0 | 0 | 0 | 38 | ▇▁▁▁▁ |
| look | 0 | 1 | 0.06 | 0.29 | 0 | 0 | 0 | 0 | 10 | ▇▁▁▁▁ |
| purchased | 0 | 1 | 0.08 | 0.37 | 0 | 0 | 0 | 0 | 13 | ▇▁▁▁▁ |
| care | 0 | 1 | 0.08 | 0.42 | 0 | 0 | 0 | 0 | 18 | ▇▁▁▁▁ |
| owed | 0 | 1 | 0.08 | 0.38 | 0 | 0 | 0 | 0 | 21 | ▇▁▁▁▁ |
| manager | 0 | 1 | 0.11 | 0.53 | 0 | 0 | 0 | 0 | 16 | ▇▁▁▁▁ |
| speak | 0 | 1 | 0.11 | 0.44 | 0 | 0 | 0 | 0 | 15 | ▇▁▁▁▁ |
| showing | 0 | 1 | 0.08 | 0.37 | 0 | 0 | 0 | 0 | 14 | ▇▁▁▁▁ |
Machine Learning Step #1: Data splitting
Split into training and testing dataset:
set.seed(9876)
split <- initial_split(matrix_df, strata=Product, prop=2/3)
training <- training(split)
testing <- testing(split)
count(training) # 60649 records## n
## 1 60649count(testing) # 30326 records## n
## 1 30326# approximately 2:1, as we specified Create 20 folds for training:
set.seed(9876)
vfold_train <- rsample::vfold_cv(data = training, v = 20)
vfold_train## # 20-fold cross-validation
## # A tibble: 20 × 2
## splits id
## <list> <chr>
## 1 <split [57616/3033]> Fold01
## 2 <split [57616/3033]> Fold02
## 3 <split [57616/3033]> Fold03
## 4 <split [57616/3033]> Fold04
## 5 <split [57616/3033]> Fold05
## 6 <split [57616/3033]> Fold06
## 7 <split [57616/3033]> Fold07
## 8 <split [57616/3033]> Fold08
## 9 <split [57616/3033]> Fold09
## 10 <split [57617/3032]> Fold10
## 11 <split [57617/3032]> Fold11
## 12 <split [57617/3032]> Fold12
## 13 <split [57617/3032]> Fold13
## 14 <split [57617/3032]> Fold14
## 15 <split [57617/3032]> Fold15
## 16 <split [57617/3032]> Fold16
## 17 <split [57617/3032]> Fold17
## 18 <split [57617/3032]> Fold18
## 19 <split [57617/3032]> Fold19
## 20 <split [57617/3032]> Fold20pull(vfold_train, splits)## [[1]]
## <Analysis/Assess/Total>
## <57616/3033/60649>
##
## [[2]]
## <Analysis/Assess/Total>
## <57616/3033/60649>
##
## [[3]]
## <Analysis/Assess/Total>
## <57616/3033/60649>
##
## [[4]]
## <Analysis/Assess/Total>
## <57616/3033/60649>
##
## [[5]]
## <Analysis/Assess/Total>
## <57616/3033/60649>
##
## [[6]]
## <Analysis/Assess/Total>
## <57616/3033/60649>
##
## [[7]]
## <Analysis/Assess/Total>
## <57616/3033/60649>
##
## [[8]]
## <Analysis/Assess/Total>
## <57616/3033/60649>
##
## [[9]]
## <Analysis/Assess/Total>
## <57616/3033/60649>
##
## [[10]]
## <Analysis/Assess/Total>
## <57617/3032/60649>
##
## [[11]]
## <Analysis/Assess/Total>
## <57617/3032/60649>
##
## [[12]]
## <Analysis/Assess/Total>
## <57617/3032/60649>
##
## [[13]]
## <Analysis/Assess/Total>
## <57617/3032/60649>
##
## [[14]]
## <Analysis/Assess/Total>
## <57617/3032/60649>
##
## [[15]]
## <Analysis/Assess/Total>
## <57617/3032/60649>
##
## [[16]]
## <Analysis/Assess/Total>
## <57617/3032/60649>
##
## [[17]]
## <Analysis/Assess/Total>
## <57617/3032/60649>
##
## [[18]]
## <Analysis/Assess/Total>
## <57617/3032/60649>
##
## [[19]]
## <Analysis/Assess/Total>
## <57617/3032/60649>
##
## [[20]]
## <Analysis/Assess/Total>
## <57617/3032/60649>Creating a recipe, selecting a model
Specify variables with the recipe() function
Create a recipe:
recipe <- training %>% recipe(Product ~ .)
# the . notation indicates that the rest of the variables are predictorsCreate a model: decision tree
model <- decision_tree() %>%
set_mode("classification") %>%
set_engine("rpart")
model## Decision Tree Model Specification (classification)
##
## Computational engine: rpartCreate a workflow:
workflow <- workflow() %>%
add_recipe(recipe) %>%
add_model(model)
workflow## ══ Workflow ════════════════════════════════════════════════════════════════════
## Preprocessor: Recipe
## Model: decision_tree()
##
## ── Preprocessor ────────────────────────────────────────────────────────────────
## 0 Recipe Steps
##
## ── Model ───────────────────────────────────────────────────────────────────────
## Decision Tree Model Specification (classification)
##
## Computational engine: rpartModel fit: without cross validation
fit <- fit(workflow, data=training)
workflow_fit<-fit %>%
pull_workflow_fit()## Warning: `pull_workflow_fit()` was deprecated in workflows 0.2.3.
## ℹ Please use `extract_fit_parsnip()` instead.workflow_fit## parsnip model object
##
## n= 60649
##
## node), split, n, loss, yval, (yprob)
## * denotes terminal node
##
## 1) root 60649 35120 Credit card or prepaid card (0.420930271 0.340285907 0.137232271 0.101551551)
## 2) mortgage< 0.5 45394 20220 Credit card or prepaid card (0.554566683 0.131999824 0.179803498 0.133629995)
## 4) card>=0.5 18430 527 Credit card or prepaid card (0.971405317 0.005154639 0.011448725 0.011991319) *
## 5) card< 0.5 26964 19013 Student loan (0.269655837 0.218699006 0.294874648 0.216770509)
## 10) student< 0.5 23189 15941 Credit card or prepaid card (0.312561991 0.253697874 0.182586571 0.251153564)
## 20) car< 0.5 20234 13097 Credit card or prepaid card (0.352723139 0.288030048 0.206385292 0.152861520)
## 40) loan< 0.5 12434 5672 Credit card or prepaid card (0.543831430 0.195110182 0.131413865 0.129644523)
## 80) loans< 0.5 11725 4978 Credit card or prepaid card (0.575437100 0.197697228 0.091940299 0.134925373)
## 160) escrow< 0.5 11319 4579 Credit card or prepaid card (0.595458963 0.169714639 0.095238095 0.139588303) *
## 161) escrow>=0.5 406 9 Mortgage (0.017241379 0.977832512 0.000000000 0.004926108) *
## 81) loans>=0.5 709 153 Student loan (0.021156559 0.152327221 0.784203103 0.042313117) *
## 41) loan>=0.5 7800 4398 Mortgage (0.048076923 0.436153846 0.325897436 0.189871795)
## 82) home>=0.5 1534 172 Mortgage (0.012385919 0.887874837 0.046936115 0.052803129) *
## 83) home< 0.5 6266 3796 Student loan (0.056814555 0.325566550 0.394190871 0.223428024)
## 166) loans< 0.5 4986 3082 Mortgage (0.070196550 0.381869234 0.285800241 0.262133975) *
## 167) loans>=0.5 1280 235 Student loan (0.004687500 0.106250000 0.816406250 0.072656250) *
## 21) car>=0.5 2955 224 Vehicle loan or lease (0.037563452 0.018612521 0.019627750 0.924196277) *
## 11) student>=0.5 3775 58 Student loan (0.006092715 0.003708609 0.984635762 0.005562914) *
## 3) mortgage>=0.5 15255 609 Mortgage (0.023271059 0.960078663 0.010553917 0.006096362) *See which variables are important:
workflow_fit$fit$variable.importance## mortgage card loan student home escrow
## 1.156521e+04 7.228772e+03 2.931520e+03 2.792036e+03 2.213974e+03 2.145142e+03
## car credit property loans modification foreclosure
## 2.067258e+03 1.989351e+03 1.680765e+03 1.608343e+03 1.402037e+03 1.358019e+03
## charges charge cards servicing application program
## 8.585883e+02 8.327012e+02 8.260333e+02 1.277481e+02 1.084374e+02 2.736204e+01
## years federal closing making got close
## 2.313052e+01 1.288626e+01 1.127550e+01 8.875347e+00 7.695376e+00 6.765302e+00
## house sold payments issues back plan
## 6.765302e+00 4.197478e+00 4.034963e+00 3.497898e+00 2.798318e+00 1.898763e+00
## form
## 7.516392e-01We see such terms as mortgage, card, loan, student, home, escrow, car, credit are among the most important
Perform a prediction:
predicted <- predict(workflow_fit, new_data = training)
accuracy(training, truth = Product, estimate = predicted$.pred_class)## # A tibble: 1 × 3
## .metric .estimator .estimate
## <chr> <chr> <dbl>
## 1 accuracy multiclass 0.841Model accuracy is 0.841 -> 84.1%
Compare counts of Product vs. predicted Product:
count(training, Product)## Product n
## 1 Credit card or prepaid card 25529
## 2 Mortgage 20638
## 3 Student loan 8323
## 4 Vehicle loan or lease 6159count(predicted, .pred_class)## # A tibble: 4 × 2
## .pred_class n
## <fct> <int>
## 1 Credit card or prepaid card 29749
## 2 Mortgage 22181
## 3 Student loan 5764
## 4 Vehicle loan or lease 2955Compare by row:
predicted_vs_true <- bind_cols(training, predicted_Product=pull(predicted, .pred_class)) %>%
select(predicted_Product, everything())
head(predicted_vs_true)## predicted_Product Product account amount bank
## 6 Credit card or prepaid card Credit card or prepaid card 3 0 0
## 8 Credit card or prepaid card Credit card or prepaid card 1 0 0
## 12 Credit card or prepaid card Credit card or prepaid card 0 0 0
## 16 Credit card or prepaid card Credit card or prepaid card 2 0 0
## 17 Credit card or prepaid card Credit card or prepaid card 1 0 0
## 18 Credit card or prepaid card Credit card or prepaid card 3 1 3
## card case charge charged charges chase closed feel fraud getting immediately
## 6 2 0 0 0 0 0 0 0 0 0 1
## 8 0 0 0 0 1 0 0 0 0 0 0
## 12 4 0 0 0 0 0 0 0 1 1 0
## 16 0 1 0 2 0 0 0 0 0 0 0
## 17 7 0 0 0 0 0 0 0 1 0 1
## 18 5 1 1 0 0 0 4 0 0 0 1
## interest last late later letter make many months new original pay sent
## 6 0 0 0 0 0 0 0 0 0 0 0 0
## 8 0 0 0 0 0 0 0 0 0 0 0 0
## 12 0 0 0 1 0 0 1 1 0 1 1 0
## 16 7 0 2 1 0 0 0 2 0 0 0 0
## 17 0 0 0 0 0 0 0 0 4 0 1 0
## 18 0 0 0 0 2 0 0 0 0 0 0 1
## statement stating ago asked back contact contacted credit debt dispute ever
## 6 0 0 0 0 0 0 0 2 0 0 0
## 8 0 0 0 0 0 0 0 1 0 0 1
## 12 0 0 0 1 1 0 0 4 0 0 0
## 16 0 0 0 0 1 0 0 0 0 0 0
## 17 0 0 0 0 0 0 0 2 0 0 0
## 18 0 0 0 0 0 0 0 1 0 0 0
## financial first form given going got happened home income mortgage note now
## 6 0 0 0 0 0 0 0 0 0 0 0 1
## 8 0 0 0 0 0 0 0 0 0 0 0 0
## 12 0 0 0 0 0 1 0 0 0 0 0 1
## 16 0 0 0 0 0 0 0 0 0 0 0 0
## 17 0 0 0 0 0 0 0 1 0 0 0 0
## 18 0 1 0 0 1 0 0 0 0 0 0 0
## one order owe past period point receive received refused removed report
## 6 0 0 0 2 0 0 1 0 0 0 0
## 8 0 0 0 0 0 0 0 0 0 0 0
## 12 0 0 0 1 0 0 0 1 0 0 0
## 16 0 0 0 0 0 0 0 0 0 0 0
## 17 0 1 0 0 0 0 0 0 0 0 1
## 18 0 0 0 0 0 0 0 1 0 0 0
## request requesting send spoke stated submitted time told using way working
## 6 0 0 0 0 0 0 1 0 0 0 0
## 8 0 0 0 0 0 0 0 0 0 0 0
## 12 0 0 0 0 0 0 0 1 0 1 0
## 16 1 0 0 0 0 0 1 2 0 0 0
## 17 0 0 0 0 0 0 0 0 0 0 0
## 18 0 0 0 1 0 0 1 4 1 0 0
## year years gave information like loans making put student transferred
## 6 0 0 0 0 0 0 0 0 1 0
## 8 0 0 0 0 0 0 0 0 0 0
## 12 0 0 0 1 0 0 0 0 0 0
## 16 0 1 0 0 0 0 0 0 0 0
## 17 0 0 0 1 0 0 0 0 0 0
## 18 0 0 0 0 1 0 0 1 0 0
## business service federal instead review second take fees full keep made mail
## 6 0 0 0 0 0 0 0 1 1 1 2 1
## 8 0 0 0 0 0 0 0 0 0 0 0 0
## 12 0 0 0 0 0 0 1 0 0 0 0 0
## 16 0 0 0 0 2 0 0 0 0 0 0 0
## 17 0 0 0 0 0 0 0 2 0 0 2 1
## 18 0 0 0 0 0 0 0 0 0 0 1 0
## online paid payment purchase score system thought work answer applied called
## 6 1 3 2 3 1 1 2 1 0 0 0
## 8 0 0 0 0 0 0 0 0 0 0 1
## 12 0 3 2 0 0 0 0 0 0 0 2
## 16 0 1 2 0 0 0 0 0 0 3 1
## 17 1 0 0 2 0 0 0 0 0 0 1
## 18 0 0 0 0 0 0 0 0 0 0 1
## check fee previous provided representative response started three calls
## 6 0 0 0 0 0 0 0 0 0
## 8 0 0 0 0 0 0 0 0 1
## 12 0 0 0 0 0 0 0 0 0
## 16 0 0 0 0 0 1 0 0 1
## 17 1 0 0 0 0 0 0 1 0
## 18 0 0 0 0 0 0 0 0 0
## company days end hours payments phone agreement asking loan office paperwork
## 6 0 0 0 0 0 0 0 0 0 0 0
## 8 1 1 1 1 1 2 0 0 0 0 0
## 12 1 0 0 0 4 0 0 0 0 0 0
## 16 0 1 0 0 0 0 0 0 0 0 0
## 17 0 0 0 0 0 0 0 0 0 0 0
## 18 0 0 0 0 0 0 0 0 0 0 0
## paying regarding statements understand advised balance money monthly rep
## 6 0 0 0 0 0 0 0 0 0
## 8 0 0 0 0 0 0 0 0 0
## 12 0 0 0 0 0 1 0 0 0
## 16 0 0 0 0 0 0 0 0 0
## 17 0 0 0 0 0 0 1 0 0
## 18 0 0 0 0 0 0 0 0 0
## reporting returned website week complete department documents modification
## 6 0 0 0 0 0 0 0 0
## 8 0 0 0 0 0 0 0 0
## 12 0 0 0 0 0 0 0 0
## 16 0 0 0 0 0 0 0 0
## 17 0 0 0 0 0 0 0 0
## 18 0 0 0 0 0 0 0 0
## month number process stop funds give matter provide say thank times
## 6 0 0 0 0 0 0 0 0 0 0 0
## 8 0 0 0 0 0 0 0 0 0 0 0
## 12 0 0 1 0 1 1 1 1 1 1 1
## 16 0 0 0 0 0 0 0 0 0 0 0
## 17 0 0 0 0 0 1 0 0 0 0 0
## 18 0 0 0 0 0 0 1 0 0 0 0
## transaction believe currently program services several address calling claim
## 6 0 0 0 0 0 0 0 0 0
## 8 0 0 0 0 0 0 0 0 0
## 12 1 0 0 0 0 0 0 0 0
## 16 0 0 1 0 0 0 0 0 0
## 17 0 0 0 0 0 0 2 0 0
## 18 0 1 0 0 0 0 0 0 0
## filed informed let someone status used continue found taken based cfpb rate
## 6 0 0 0 0 0 0 0 0 0 0 0 0
## 8 0 0 0 0 0 0 0 0 0 0 0 0
## 12 0 0 0 0 0 0 0 0 0 0 0 0
## 16 0 0 0 0 0 0 0 0 0 1 1 6
## 17 0 1 0 0 1 0 0 0 0 0 0 0
## 18 0 0 1 0 0 1 0 0 0 0 0 0
## refund requested resolve total tried change changed day notified wrong close
## 6 0 0 0 0 0 0 0 0 0 0 0
## 8 0 0 0 0 0 0 0 0 0 0 0
## 12 0 0 0 0 0 0 0 0 0 0 0
## 16 1 1 1 1 1 0 0 0 0 0 0
## 17 0 0 0 0 0 1 1 1 2 1 0
## 18 0 0 0 0 0 0 0 0 0 0 2
## offered opened bill car hold reason said agent fraudulent insurance property
## 6 0 0 0 0 0 0 0 0 0 0 0
## 8 0 0 0 0 0 0 0 0 0 0 0
## 12 0 0 0 0 0 0 0 0 0 0 0
## 16 0 0 0 0 0 0 0 0 0 0 0
## 17 0 0 0 0 0 0 0 0 0 0 0
## 18 1 1 0 0 0 0 0 0 0 0 0
## servicing already correct current dated foreclosure law part place required
## 6 0 0 0 0 0 0 0 0 0 0
## 8 0 0 0 0 0 0 0 0 0 0
## 12 0 0 0 0 0 0 0 0 0 0
## 16 0 0 0 0 0 0 0 0 0 0
## 17 0 0 0 0 0 0 0 0 0 0
## 18 0 0 0 0 0 0 0 0 0 0
## came house lender application apply needed plan situation customer long
## 6 0 0 0 0 0 0 0 0 0 0
## 8 0 0 0 0 0 0 0 0 0 0
## 12 0 0 0 0 0 0 0 0 0 0
## 16 0 0 0 0 0 0 0 0 0 0
## 17 0 0 0 0 0 0 0 0 0 0
## 18 0 0 0 0 0 0 0 0 0 0
## multiple saying terms weeks attached available denied email error set
## 6 0 0 0 0 0 0 0 0 0 0
## 8 0 0 0 0 0 0 0 0 0 0
## 12 0 0 0 0 0 0 0 0 0 0
## 16 0 0 0 0 0 0 0 0 0 0
## 17 0 0 0 0 0 0 0 0 0 0
## 18 0 0 0 0 0 0 0 0 0 0
## closing sold tell escrow took attorney ive resolved reported state states
## 6 0 0 0 0 0 0 0 0 0 0 0
## 8 0 0 0 0 0 0 0 0 0 0 0
## 12 0 0 0 0 0 0 0 0 0 0 0
## 16 0 0 0 0 0 0 0 0 0 0 0
## 17 0 0 0 0 0 0 0 0 0 0 0
## 18 0 0 0 0 0 0 0 0 0 0 0
## file though two different longer receiving accounts consumer documentation
## 6 0 0 0 0 0 0 0 0 0
## 8 0 0 0 0 0 0 0 0 0
## 12 0 0 0 0 0 0 0 0 0
## 16 0 0 0 0 0 0 0 0 0
## 17 0 0 0 0 0 0 0 0 0
## 18 0 0 0 0 0 0 0 0 0
## nothing return cards line notice signed explained proof approved offer
## 6 0 0 0 0 0 0 0 0 0 0
## 8 0 0 0 0 0 0 0 0 0 0
## 12 0 0 0 0 0 0 0 0 0 0
## 16 0 0 0 0 0 0 0 0 0 0
## 17 0 0 0 0 0 0 0 0 0 0
## 18 0 0 0 0 0 0 0 0 0 0
## supervisor trying agreed copy issues history show transfer look purchased
## 6 0 0 0 0 0 0 0 0 0 0
## 8 0 0 0 0 0 0 0 0 0 0
## 12 0 0 0 0 0 0 0 0 0 0
## 16 0 0 0 0 0 0 0 0 0 0
## 17 0 0 0 0 0 0 0 0 0 0
## 18 0 0 0 0 0 0 0 0 0 0
## care owed manager speak showing
## 6 0 0 0 0 0
## 8 0 0 0 0 0
## 12 0 0 0 0 0
## 16 0 0 0 0 0
## 17 0 0 0 0 0
## 18 0 0 0 0 0Visualize: side-by-side barchart
ggplot(data = predicted_vs_true, aes(x = predicted_Product, fill = Product)) +
geom_bar(position = "dodge") +
labs(x = "Predicted Product", y = "Frequency")
Model fit: with cross validation
resample_fit <- fit_resamples(workflow, vfold_train)
collect_metrics(resample_fit)## # A tibble: 2 × 6
## .metric .estimator mean n std_err .config
## <chr> <chr> <dbl> <int> <dbl> <chr>
## 1 accuracy multiclass 0.841 20 0.000866 Preprocessor1_Model1
## 2 roc_auc hand_till 0.930 20 0.000982 Preprocessor1_Model1Accuracy is still 0.841 -> 84.1%
Tuning hyperparameters: can improve model performance
Complexity parameter = cost_complexity(); depth parameter = tree_depth()
model_tune <- parsnip::decision_tree(cost_complexity=tune(), tree_depth=tune()) %>%
parsnip::set_mode("classification") %>%
parsnip::set_engine("rpart")
model_tune## Decision Tree Model Specification (classification)
##
## Main Arguments:
## cost_complexity = tune()
## tree_depth = tune()
##
## Computational engine: rpartNew workflow:
workflow_tune <- workflow() %>%
add_recipe(recipe) %>%
add_model(model_tune)Fit a model:
set.seed(1234)
tune_grid <- grid_regular(cost_complexity(), tree_depth(), levels=3)
resample_fit2 <- tune_grid(workflow_tune,
resamples = vfold_train,
grid = tune_grid,
metrics=metric_set(accuracy, roc_auc))Assess performance:
See the min_n values for the top performing models (those with the highest accuracy)
show_best(resample_fit2, metric="accuracy")## # A tibble: 5 × 8
## cost_complexity tree_depth .metric .estimator mean n std_err .config
## <dbl> <int> <chr> <chr> <dbl> <int> <dbl> <chr>
## 1 0.0000000001 15 accuracy multiclass 0.879 20 0.00113 Preproces…
## 2 0.00000316 15 accuracy multiclass 0.879 20 0.00113 Preproces…
## 3 0.0000000001 8 accuracy multiclass 0.854 20 0.000805 Preproces…
## 4 0.00000316 8 accuracy multiclass 0.854 20 0.000805 Preproces…
## 5 0.0000000001 1 accuracy multiclass 0.657 20 0.00138 Preproces…tuned_values <- select_best(resample_fit2, "accuracy")
tuned_values## # A tibble: 1 × 3
## cost_complexity tree_depth .config
## <dbl> <int> <chr>
## 1 0.0000000001 15 Preprocessor1_Model7workflow_tuned <- workflow_tune %>%
finalize_workflow(tuned_values)Finalize the model/workflow that we we used for tuning with these values:
?last_fit
overallfit <- last_fit(object = workflow_tuned, split = split)
collect_metrics(overallfit)## # A tibble: 2 × 4
## .metric .estimator .estimate .config
## <chr> <chr> <dbl> <chr>
## 1 accuracy multiclass 0.878 Preprocessor1_Model1
## 2 roc_auc hand_till 0.960 Preprocessor1_Model1Accuracy is 0.878 -> 87.8%
See predicted values:
predictions <- collect_predictions(overallfit)Visualize: side-by-side barchart
ggplot(data = predictions, aes(x = .pred_class, fill = Product)) +
geom_bar(position = "dodge") +
labs(x = "Predicted Product", y = "Frequency")
Alternative: stacked barchart
ggplot(data = predictions, aes(x = .pred_class, fill = Product)) +
geom_bar() +
labs(x = "Predicted Product", y = "Frequency")
Working with testing data file
Look at the testing data:
head(test)## problem_id
## 1 1
## 2 2
## 3 3
## 4 4
## 5 5
## 6 6
## Consumer.complaint.narrative
## 1 I have multiple lateness/missed payments on my credit report due to ECMC not properly contacting me to inform me when payments were due. I never received any statements in regards to my loan. It was also a period in time when my loan was deferred so there should not be any late/ missed payments that is being reporting on my credit file.
## 2 I lost my job in XXXX due to the coronavirus and was not able to get unemployment I been filing and calling and appealing my claim but still no money I lost my house and staying with my parents I told this company that and they have a device on the car to stop it with no payment. Do they stop my car I told them that I finally found a temporary job its temp to hire and it starts this Monday XX/XX/XXXX I sent the email and ask please send a signal to my call so I can go sign my paperwork to start they told me Im behind on my note and that they would not be able to send the signal unless I pay XXXX dollars I cried on the phone begging and pleading eith them I told them I didnt have it that my family has helped me since XXXX I can give them XXXX dollars and with my fist check in XXXX I could pay XXXX every two weeks they told me no and the guy XXXX was very rude abs belittling and I ask to speak to the manager XXXX and he told me she was in a meeting and that the information came from her I cried I said please let me talk to her he told me she would call me after the meeting its now XXXX on Friday XX/XX/XXXX I guess shes still in the meeting this is the worst company to get a car from I will never do business with them again I told them to come get the car I have been through too much this year I lost my XXXX to the XXXX and I lost my XXXX to XXXX/ the XXXX I lost my house my job and now my car
## 3 I had contacted great lakes about my student loan. I told them I couldn't afford payments i couldn't find a job. So they put me in forberance but didnt quit explaining. It to me this was the end of XXXX. Well when I got on the pay as you go plan I got charged the interest than for some reason there was. Months were they put me in forberance tha. The same month on a payment plan and that went back and fourth like that for quite a few months. I turned in my paperwork at the end of XXXX but they didn't approve it till XXXX so I got charged interest. As well as XXXX XXXX to them about the borrowers defense they shut me down didnt even try to help. They know I work for a non profit organization and am gon na try for the pslf but they have never said that they don't dont do the qualifying for that as well as I've talked several times with the csr and wanted to make sure I was on the right payment plan to count for the pslf loan they would tell me one thing but than it would change I still dont know exactly if the pmt plan im on is correct i got nothing but resistance when trying the borrowers defense loan. The company is not very good and i dont feel that any advice they give is in to help me out. Just their pockets I learned that early on dealing with them
## 4 Wells Fargo Home Mortgage put me in a payment suspension program 3 months ago, even though I did not need it. I was not suffering a hardship from Covid 19. I not only haven't missed a payment during the 3-month forbearance, I've actually made 3 additional mortgage payments in that time period. On the Wells Fargo website, it is stated that I can end the payment suspension at any time. THAT IS A LIE. I have requested that the program end multiple times, and they still have not done it. In addition, because I am still in forbearance, my mortgage account statements are not accurate or transparent. The information on my statements appear to be based on a strange and questionable accounting program devised by Wells Fargo. What credible reason could Wells Fargo possibly have for their obscure accounting, as statements should be accurate at all times. I have questioned WF executives about their accounting and I've been told repeatedly that the statements will be adjusted at a later time. What kind of BS is that? I am requesting again, that I be taken out of the payment suspension program. I also expect Wells Fargo to provide me with an updated and ACCURATE mortgage statement. It doesn't say much for the credibility of Wells Fargo Home Mortgage that I have to address these issues through Consumer Financial Protection Bureau..
## 5 We had a loan through Kia Motors Finance on our XXXX XXXX which we paid off XX/XX/XXXX. Since then, we have called customer service numerous times checking the status of our overpayment refund, but each time the story is the same : that a refund has been processed but not sent, and that there is no way to contact the responsible department other than by email, to which there has to-date never been a response. I have been reassured by various reps that they are looking into it, but nothing has changed. \n\nIt is now more than XXXX months later. If I were XXXX months late with a payment, Kia would at least continue charging interest, if not take legal action. I have contacted Kia 's complaint department in writing ( the only method which I was told will guarantee a response ), and I will be taking legal action if this matter is not addressed shortly. The pandemic is no excuse. Everyone else has carried on business as usual. Kia has a responsibility to as well.
## 6 I have received a billing statement from HOME DEPOT for a credit card which has NEVER been applied by me, this was received in USPS mail on XX/XX/2020. \n\nThe amount of credit limit on the statement shows up as {$8000.00} and the Purchases made fraudulently on that are of {$5300.00} I had called the HOME DEPOT credit card Customer Service on XXXX, and they were not able to resolve so they connected me to their fraud protection agency who were on the phone with me on XX/XX/2020 for 1 hour 16 minutes from XXXX CT Towards the end of my call with them, the fraud protection agency did confirm that the issue has been resolved and provided a reference # XXXX, but I repeatedly asked them to send me some communication by mail or email, but they were not willing to do so. \n\nI just need an assurance that this is resolved and that I am not liable to pay any amounts for purchases which have NOT been made.
## Company State ZIP.code Submitted.via
## 1 ECMC Group, Inc. NY 112XX Web
## 2 Michael Wayne Investment VA 234XX Web
## 3 Nelnet, Inc. AR 727XX Web
## 4 WELLS FARGO & COMPANY PA 189XX Web
## 5 HYUNDAI CAPITAL AMERICA IN 471XX Web
## 6 CITIBANK, N.A. TX 750XX WebClean the data similarly to train data:
test$Consumer.complaint.narrative <- gsub("XX/", "", test$Consumer.complaint.narrative)
test$Consumer.complaint.narrative <- gsub("X/", "", test$Consumer.complaint.narrative)
test$Consumer.complaint.narrative <- gsub("[XX]", "", test$Consumer.complaint.narrative)
test$Consumer.complaint.narrative <- gsub("[[:punct:]]", "", test$Consumer.complaint.narrative)
test$Consumer.complaint.narrative <- gsub("[0-9]", "", test$Consumer.complaint.narrative)
test$Consumer.complaint.narrative <- str_to_lower(test$Consumer.complaint.narrative)
head(test)## problem_id
## 1 1
## 2 2
## 3 3
## 4 4
## 5 5
## 6 6
## Consumer.complaint.narrative
## 1 i have multiple latenessmissed payments on my credit report due to ecmc not properly contacting me to inform me when payments were due i never received any statements in regards to my loan it was also a period in time when my loan was deferred so there should not be any late missed payments that is being reporting on my credit file
## 2 i lost my job in due to the coronavirus and was not able to get unemployment i been filing and calling and appealing my claim but still no money i lost my house and staying with my parents i told this company that and they have a device on the car to stop it with no payment do they stop my car i told them that i finally found a temporary job its temp to hire and it starts this monday i sent the email and ask please send a signal to my call so i can go sign my paperwork to start they told me im behind on my note and that they would not be able to send the signal unless i pay dollars i cried on the phone begging and pleading eith them i told them i didnt have it that my family has helped me since i can give them dollars and with my fist check in i could pay every two weeks they told me no and the guy was very rude abs belittling and i ask to speak to the manager and he told me she was in a meeting and that the information came from her i cried i said please let me talk to her he told me she would call me after the meeting its now on friday i guess shes still in the meeting this is the worst company to get a car from i will never do business with them again i told them to come get the car i have been through too much this year i lost my to the and i lost my to the i lost my house my job and now my car
## 3 i had contacted great lakes about my student loan i told them i couldnt afford payments i couldnt find a job so they put me in forberance but didnt quit explaining it to me this was the end of well when i got on the pay as you go plan i got charged the interest than for some reason there was months were they put me in forberance tha the same month on a payment plan and that went back and fourth like that for quite a few months i turned in my paperwork at the end of but they didnt approve it till so i got charged interest as well as to them about the borrowers defense they shut me down didnt even try to help they know i work for a non profit organization and am gon na try for the pslf but they have never said that they dont dont do the qualifying for that as well as ive talked several times with the csr and wanted to make sure i was on the right payment plan to count for the pslf loan they would tell me one thing but than it would change i still dont know exactly if the pmt plan im on is correct i got nothing but resistance when trying the borrowers defense loan the company is not very good and i dont feel that any advice they give is in to help me out just their pockets i learned that early on dealing with them
## 4 wells fargo home mortgage put me in a payment suspension program months ago even though i did not need it i was not suffering a hardship from covid i not only havent missed a payment during the month forbearance ive actually made additional mortgage payments in that time period on the wells fargo website it is stated that i can end the payment suspension at any time that is a lie i have requested that the program end multiple times and they still have not done it in addition because i am still in forbearance my mortgage account statements are not accurate or transparent the information on my statements appear to be based on a strange and questionable accounting program devised by wells fargo what credible reason could wells fargo possibly have for their obscure accounting as statements should be accurate at all times i have questioned wf executives about their accounting and ive been told repeatedly that the statements will be adjusted at a later time what kind of bs is that i am requesting again that i be taken out of the payment suspension program i also expect wells fargo to provide me with an updated and accurate mortgage statement it doesnt say much for the credibility of wells fargo home mortgage that i have to address these issues through consumer financial protection bureau
## 5 we had a loan through kia motors finance on our which we paid off since then we have called customer service numerous times checking the status of our overpayment refund but each time the story is the same that a refund has been processed but not sent and that there is no way to contact the responsible department other than by email to which there has todate never been a response i have been reassured by various reps that they are looking into it but nothing has changed \n\nit is now more than months later if i were months late with a payment kia would at least continue charging interest if not take legal action i have contacted kia s complaint department in writing the only method which i was told will guarantee a response and i will be taking legal action if this matter is not addressed shortly the pandemic is no excuse everyone else has carried on business as usual kia has a responsibility to as well
## 6 i have received a billing statement from home depot for a credit card which has never been applied by me this was received in usps mail on \n\nthe amount of credit limit on the statement shows up as and the purchases made fraudulently on that are of i had called the home depot credit card customer service on and they were not able to resolve so they connected me to their fraud protection agency who were on the phone with me on for hour minutes from ct towards the end of my call with them the fraud protection agency did confirm that the issue has been resolved and provided a reference but i repeatedly asked them to send me some communication by mail or email but they were not willing to do so \n\ni just need an assurance that this is resolved and that i am not liable to pay any amounts for purchases which have not been made
## Company State ZIP.code Submitted.via
## 1 ECMC Group, Inc. NY 112XX Web
## 2 Michael Wayne Investment VA 234XX Web
## 3 Nelnet, Inc. AR 727XX Web
## 4 WELLS FARGO & COMPANY PA 189XX Web
## 5 HYUNDAI CAPITAL AMERICA IN 471XX Web
## 6 CITIBANK, N.A. TX 750XX WebCreate a term matrix:
test_matrix <- DocumentTermMatrix(test$Consumer.complaint.narrative)
inspect(test_matrix)## <<DocumentTermMatrix (documents: 20, terms: 1234)>>
## Non-/sparse entries: 2350/22330
## Sparsity : 90%
## Maximal term length: 16
## Weighting : term frequency (tf)
## Sample :
## Terms
## Docs and for forbearance have not that the they was with
## 11 4 5 0 2 4 5 6 3 0 0
## 12 16 2 0 1 7 4 39 1 12 10
## 13 17 6 0 3 6 4 19 8 5 3
## 17 29 11 7 4 4 25 33 0 29 5
## 2 16 0 0 3 2 5 14 5 3 4
## 3 5 6 0 1 1 6 14 9 3 2
## 4 4 2 2 5 5 8 8 1 1 1
## 5 2 0 0 3 3 3 6 1 1 1
## 7 9 4 0 0 5 1 10 3 4 2
## 9 33 31 33 9 7 38 81 6 30 21Convert test matrix to dataframe:
test_matrix_df <- test_matrix %>%
as.matrix() %>%
as.data.frame() Make prediction:
predict_model <- fit(workflow_tuned, matrix_df)
predict(predict_model, new_data = test_matrix_df)## Error in `validate_column_names()`:
## ! The following required columns are missing: 'chase', 'happened', 'past', 'second', 'keep', 'system', 'thought', 'answer', 'previous', 'office', 'rep', 'services', 'cfpb', 'wrong', 'opened', 'insurance', 'property', 'foreclosure', 'part', 'lender', 'application', 'apply', 'long', 'sold', 'attorney', 'state', 'accounts', 'return', 'proof', 'supervisor', 'show', 'purchased', 'owed', 'showing'.We have an error: some columns that exist in training data are missing from testing data
Add them manually:
test_matrix_df$chase <- 0
test_matrix_df$cant <- 0
test_matrix_df$happened <- 0
test_matrix_df$past <- 0
test_matrix_df$something <- 0
test_matrix_df$second <- 0
test_matrix_df$keep <- 0
test_matrix_df$system <- 0
test_matrix_df$thought <- 0
test_matrix_df$answer <- 0
test_matrix_df$previous <- 0
test_matrix_df$office <- 0
test_matrix_df$rep <- 0
test_matrix_df$assistance <- 0
test_matrix_df$services <- 0
test_matrix_df$either <- 0
test_matrix_df$cfpb <- 0
test_matrix_df$wrong <- 0
test_matrix_df$opened <- 0
test_matrix_df$insurance <- 0
test_matrix_df$property <- 0
test_matrix_df$foreclosure <- 0
test_matrix_df$part <- 0
test_matrix_df$lender <- 0
test_matrix_df$application <- 0
test_matrix_df$apply <- 0
test_matrix_df$recently <- 0
test_matrix_df$always <- 0
test_matrix_df$long <- 0
test_matrix_df$sold <- 0
test_matrix_df$attorney <- 0
test_matrix_df$state <- 0
test_matrix_df$must <- 0
test_matrix_df$accounts <- 0
test_matrix_df$return <- 0
test_matrix_df$proof <- 0
test_matrix_df$supervisor <- 0
test_matrix_df$wasnt <- 0
test_matrix_df$show <- 0
test_matrix_df$purchased <- 0
test_matrix_df$person <- 0
test_matrix_df$owed <- 0
test_matrix_df$showing <- 0Make new prediction:
prediction <- predict(predict_model, new_data = test_matrix_df)Merge back to test dataset:
prediction <- bind_cols(test, predicted_Product=pull(prediction, .pred_class)) %>%
select(predicted_Product, everything())
head(prediction)## predicted_Product problem_id
## 1 Vehicle loan or lease 1
## 2 Vehicle loan or lease 2
## 3 Student loan 3
## 4 Mortgage 4
## 5 Student loan 5
## 6 Credit card or prepaid card 6
## Consumer.complaint.narrative
## 1 i have multiple latenessmissed payments on my credit report due to ecmc not properly contacting me to inform me when payments were due i never received any statements in regards to my loan it was also a period in time when my loan was deferred so there should not be any late missed payments that is being reporting on my credit file
## 2 i lost my job in due to the coronavirus and was not able to get unemployment i been filing and calling and appealing my claim but still no money i lost my house and staying with my parents i told this company that and they have a device on the car to stop it with no payment do they stop my car i told them that i finally found a temporary job its temp to hire and it starts this monday i sent the email and ask please send a signal to my call so i can go sign my paperwork to start they told me im behind on my note and that they would not be able to send the signal unless i pay dollars i cried on the phone begging and pleading eith them i told them i didnt have it that my family has helped me since i can give them dollars and with my fist check in i could pay every two weeks they told me no and the guy was very rude abs belittling and i ask to speak to the manager and he told me she was in a meeting and that the information came from her i cried i said please let me talk to her he told me she would call me after the meeting its now on friday i guess shes still in the meeting this is the worst company to get a car from i will never do business with them again i told them to come get the car i have been through too much this year i lost my to the and i lost my to the i lost my house my job and now my car
## 3 i had contacted great lakes about my student loan i told them i couldnt afford payments i couldnt find a job so they put me in forberance but didnt quit explaining it to me this was the end of well when i got on the pay as you go plan i got charged the interest than for some reason there was months were they put me in forberance tha the same month on a payment plan and that went back and fourth like that for quite a few months i turned in my paperwork at the end of but they didnt approve it till so i got charged interest as well as to them about the borrowers defense they shut me down didnt even try to help they know i work for a non profit organization and am gon na try for the pslf but they have never said that they dont dont do the qualifying for that as well as ive talked several times with the csr and wanted to make sure i was on the right payment plan to count for the pslf loan they would tell me one thing but than it would change i still dont know exactly if the pmt plan im on is correct i got nothing but resistance when trying the borrowers defense loan the company is not very good and i dont feel that any advice they give is in to help me out just their pockets i learned that early on dealing with them
## 4 wells fargo home mortgage put me in a payment suspension program months ago even though i did not need it i was not suffering a hardship from covid i not only havent missed a payment during the month forbearance ive actually made additional mortgage payments in that time period on the wells fargo website it is stated that i can end the payment suspension at any time that is a lie i have requested that the program end multiple times and they still have not done it in addition because i am still in forbearance my mortgage account statements are not accurate or transparent the information on my statements appear to be based on a strange and questionable accounting program devised by wells fargo what credible reason could wells fargo possibly have for their obscure accounting as statements should be accurate at all times i have questioned wf executives about their accounting and ive been told repeatedly that the statements will be adjusted at a later time what kind of bs is that i am requesting again that i be taken out of the payment suspension program i also expect wells fargo to provide me with an updated and accurate mortgage statement it doesnt say much for the credibility of wells fargo home mortgage that i have to address these issues through consumer financial protection bureau
## 5 we had a loan through kia motors finance on our which we paid off since then we have called customer service numerous times checking the status of our overpayment refund but each time the story is the same that a refund has been processed but not sent and that there is no way to contact the responsible department other than by email to which there has todate never been a response i have been reassured by various reps that they are looking into it but nothing has changed \n\nit is now more than months later if i were months late with a payment kia would at least continue charging interest if not take legal action i have contacted kia s complaint department in writing the only method which i was told will guarantee a response and i will be taking legal action if this matter is not addressed shortly the pandemic is no excuse everyone else has carried on business as usual kia has a responsibility to as well
## 6 i have received a billing statement from home depot for a credit card which has never been applied by me this was received in usps mail on \n\nthe amount of credit limit on the statement shows up as and the purchases made fraudulently on that are of i had called the home depot credit card customer service on and they were not able to resolve so they connected me to their fraud protection agency who were on the phone with me on for hour minutes from ct towards the end of my call with them the fraud protection agency did confirm that the issue has been resolved and provided a reference but i repeatedly asked them to send me some communication by mail or email but they were not willing to do so \n\ni just need an assurance that this is resolved and that i am not liable to pay any amounts for purchases which have not been made
## Company State ZIP.code Submitted.via
## 1 ECMC Group, Inc. NY 112XX Web
## 2 Michael Wayne Investment VA 234XX Web
## 3 Nelnet, Inc. AR 727XX Web
## 4 WELLS FARGO & COMPANY PA 189XX Web
## 5 HYUNDAI CAPITAL AMERICA IN 471XX Web
## 6 CITIBANK, N.A. TX 750XX WebVisualize: scatter plot with Problem ID as y-axis and Predicted Product as x-axis
ggplot(data = prediction, aes(x = predicted_Product, y = problem_id)) +
geom_point() +
labs(x = "Predicted Product Category", y = "Problem ID") +
scale_y_continuous(breaks = seq(0, 20, by = 1)) +
theme_bw()