Pull in Dataset

bank <- read.csv("C:\\Users\\jashb\\OneDrive\\Documents\\Masters Data Science\\Spring 2025\\DATA 622\\Assignment 1\\DATA\\bank-additional\\bank-additional-full.csv", sep = ';')

EDA

Input variables:

other attributes:

12 - campaign: number of contacts performed during this campaign and for this client (numeric, includes last contact)

13 - pdays: number of days that passed by after the client was last contacted from a previous campaign (numeric; 999 means client was not previously contacted)

14 - previous: number of contacts performed before this campaign and for this client (numeric)

15 - poutcome: outcome of the previous marketing campaign (categorical: “failure”,“nonexistent”,“success”) # social and economic context attributes

16 - emp.var.rate: employment variation rate - quarterly indicator (numeric)

17 - cons.price.idx: consumer price index - monthly indicator (numeric)

18 - cons.conf.idx: consumer confidence index - monthly indicator (numeric)

19 - euribor3m: euribor 3 month rate - daily indicator (numeric)

20 - nr.employed: number of employees - quarterly indicator (numeric) Output variable (desired target):

21 - y - has the client subscribed a term deposit? (binary: “yes”,“no”)

STR and Skim to get an overall view of missingness, uniqueness, mean and standard deviation.

The independent variables within the dataset consist of integers and characters variables. The dependent variable is a simple “yes” or “no”

str(bank)

## 'data.frame':    41188 obs. of  21 variables:
##  $ age           : int  56 57 37 40 56 45 59 41 24 25 ...
##  $ job           : chr  "housemaid" "services" "services" "admin." ...
##  $ marital       : chr  "married" "married" "married" "married" ...
##  $ education     : chr  "basic.4y" "high.school" "high.school" "basic.6y" ...
##  $ default       : chr  "no" "unknown" "no" "no" ...
##  $ housing       : chr  "no" "no" "yes" "no" ...
##  $ loan          : chr  "no" "no" "no" "no" ...
##  $ contact       : chr  "telephone" "telephone" "telephone" "telephone" ...
##  $ month         : chr  "may" "may" "may" "may" ...
##  $ day_of_week   : chr  "mon" "mon" "mon" "mon" ...
##  $ duration      : int  261 149 226 151 307 198 139 217 380 50 ...
##  $ campaign      : int  1 1 1 1 1 1 1 1 1 1 ...
##  $ pdays         : int  999 999 999 999 999 999 999 999 999 999 ...
##  $ previous      : int  0 0 0 0 0 0 0 0 0 0 ...
##  $ poutcome      : chr  "nonexistent" "nonexistent" "nonexistent" "nonexistent" ...
##  $ emp.var.rate  : num  1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 ...
##  $ cons.price.idx: num  94 94 94 94 94 ...
##  $ cons.conf.idx : num  -36.4 -36.4 -36.4 -36.4 -36.4 -36.4 -36.4 -36.4 -36.4 -36.4 ...
##  $ euribor3m     : num  4.86 4.86 4.86 4.86 4.86 ...
##  $ nr.employed   : num  5191 5191 5191 5191 5191 ...
##  $ y             : chr  "no" "no" "no" "no" ...

skim(bank)

Variable type: character

Variable type: numeric

cor_mat <- cor(bank[,unlist(lapply(bank,is.numeric))])
corrplot(cor_mat, type = 'upper',
         method = "color", 
         addCoef.col = "black", 
         number.cex = 0.8, 
         tl.cex = 0.8, 
         tl.col = "black", 
         tl.srt = 45, 
         col = colorRampPalette(c("blue", "white", "grey"))(100))

print(colSums(is.na(bank)))

##            age            job        marital      education        default 
##              0              0              0              0              0 
##        housing           loan        contact          month    day_of_week 
##              0              0              0              0              0 
##       duration       campaign          pdays       previous       poutcome 
##              0              0              0              0              0 
##   emp.var.rate cons.price.idx  cons.conf.idx      euribor3m    nr.employed 
##              0              0              0              0              0 
##              y 
##              0

numeric_vars <- bank %>% select_if(is.numeric) %>% gather()

numeric_vars <- numeric_vars %>%
  group_by(key) %>% 
  mutate(ind = row_number()) %>% 
  ungroup()

for (var in unique(numeric_vars$key)) {
  p <- numeric_vars %>%
    filter(key == var) %>%
    ggplot(aes(value)) +
    geom_density() +
    labs(title = var) +
    theme(axis.text.x = element_text(angle = 45, hjust = 1))
  print(p)
}

non_numeric_vars <- bank %>% select_if(negate(is.numeric)) %>% gather()

for (var in unique(non_numeric_vars$key)) {
  p <- non_numeric_vars %>%
    filter(key == var) %>%
    ggplot(aes(x = value, fill = key)) +
    geom_bar() +
    labs(title = var) +
    theme(axis.text.x = element_text(angle = 45, hjust = 1))
  print(p)
}

numeric_vars %>% 
  ggplot(aes(x = ind, y = value,))+
  geom_point(color = 'salmon')+
  facet_wrap(~key, scales = "free")+
  theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
  labs(title = "Numeric Variables Scatter Plots", x = "Row_Num", y = "Value")

mode_calc <- function(v){
  uniqv <- unique(v)
  uniqv[which.max(tabulate(match(v,uniqv)))]
}

central_tab <- bank %>% 
  select(where(is.numeric)) %>% 
  summarise_all(list(
    Mean = ~mean(.,na.rm = T),
    Median = ~median(.,na.rm = T),
    Mode = ~mode_calc(.),
    Std= ~sd(.,na.rm = T),
    IQR = ~IQR(.,na.rm = T)
  ))

#Converting from wide to long
central_tendency_long <- central_tab %>% 
  pivot_longer(cols = everything(), names_to = "Variable", values_to = "Value") %>%
  separate(Variable, into = c("Variable", "Measure"), sep = "_") %>%
  pivot_wider(names_from = Measure, values_from = Value)

print(central_tendency_long)

## # A tibble: 10 × 6
##    Variable            Mean  Median    Mode     Std     IQR
##    <chr>              <dbl>   <dbl>   <dbl>   <dbl>   <dbl>
##  1 age              40.0      38      31     10.4    15    
##  2 duration        258.      180      85    259.    217    
##  3 campaign          2.57      2       1      2.77    2    
##  4 pdays           962.      999     999    187.      0    
##  5 previous          0.173     0       0      0.495   0    
##  6 emp.var.rate      0.0819    1.1     1.4    1.57    3.2  
##  7 cons.price.idx   93.6      93.7    94.0    0.579   0.919
##  8 cons.conf.idx   -40.5     -41.8   -36.4    4.63    6.30 
##  9 euribor3m         3.62      4.86    4.86   1.73    3.62 
## 10 nr.employed    5167.     5191    5228.    72.3   129

Algorithm Selection:

For this classification problem two algorithms make sense based on the preliminary view of the data. First. A logistic regression. The primary benefit of logistic regressions is they are easy to deploy and are easy to interpret. Logistic regression can also handle larger data sets making them more computationally efficient than some of the more complex algorithms. However this algorithm is sensitive to outliers and multicollinearity, two things this data has (especially outliers). The best option for this dataset will likely be Random Forest. Despite the longer run times brought about by higher complexity, this algorithm is better equipped to handle imbalanced data, as well as a mix of categorical character features and numerical integer features. Additionally, if the unknowns in the data were to be converted back to NA, random forest is also able to handle missing data fairly well. As a result the Random Forest will be chosen.

Pre-processing:

As mentioned above, there actually is missing data in the dataset it is just coded as ‘unknown’. So first preprocessing step is deciding what to do about this missing data. One option is recoding the data as NA and then running it through a program like MICE to impute. Another option is letting the Random Forest do its own computation for imputing. Regardless, the NA need to be addressed. While random forest doesn’t necessarily need standardization, dropping highly correlated variables that are measuring the same thing could improve model performance and cut down on run time for the classification. For example nr.employed and emp.var.rate are highly correlated because they are both measuring a change over time in employment, for this, the features could either be combined or only keep one.