Online Food Model

Introduction

The dataset contains information collected from an online food ordering platform over a period of time.

It shows Demographic Information, Location Information and Order details.

Attributes:

Demographic Information:

Age: Age of the customer.
Gender: Gender of the customer.
Marital Status: Marital status of the customer.
Occupation: Occupation of the customer.
Monthly Income: Monthly income of the customer.
Educational Qualifications: Educational qualifications of the customer.
Family Size: Number of individuals in the customer’s family.

Location Information:

Latitude: Latitude of the customer’s location.
Longitude: Longitude of the customer’s location.
Pin Code: Pin code of the customer’s location.

Order Details:

Output: Current status of the order (e.g., pending, confirmed, delivered).
Feedback: Feedback provided by the customer after receiving the order.

Load the Dataset

library(readr)
onlinefood <- read_csv("~/Downloads/CS 583/R_Project/onlinefood.csv")

## Rows: 388 Columns: 12
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (7): Gender, Marital Status, Occupation, Monthly Income, Educational Qua...
## dbl (5): Age, Family size, latitude, longitude, Pin code
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.

onlinefood

## # A tibble: 388 × 12
##      Age Gender `Marital Status` Occupation `Monthly Income`
##    <dbl> <chr>  <chr>            <chr>      <chr>           
##  1    20 Female Single           Student    No Income       
##  2    24 Female Single           Student    Below Rs.10000  
##  3    22 Male   Single           Student    Below Rs.10000  
##  4    22 Female Single           Student    No Income       
##  5    22 Male   Single           Student    Below Rs.10000  
##  6    27 Female Married          Employee   More than 50000 
##  7    22 Male   Single           Student    No Income       
##  8    24 Female Single           Student    No Income       
##  9    23 Female Single           Student    No Income       
## 10    23 Female Single           Student    No Income       
## # ℹ 378 more rows
## # ℹ 7 more variables: `Educational Qualifications` <chr>, `Family size` <dbl>,
## #   latitude <dbl>, longitude <dbl>, `Pin code` <dbl>, Output <chr>,
## #   Feedback <chr>

Analysis

dim(onlinefood)

## [1] 388  12

summary(onlinefood)

##       Age           Gender          Marital Status      Occupation       
##  Min.   :18.00   Length:388         Length:388         Length:388        
##  1st Qu.:23.00   Class :character   Class :character   Class :character  
##  Median :24.00   Mode  :character   Mode  :character   Mode  :character  
##  Mean   :24.63                                                           
##  3rd Qu.:26.00                                                           
##  Max.   :33.00                                                           
##  Monthly Income     Educational Qualifications  Family size       latitude    
##  Length:388         Length:388                 Min.   :1.000   Min.   :12.87  
##  Class :character   Class :character           1st Qu.:2.000   1st Qu.:12.94  
##  Mode  :character   Mode  :character           Median :3.000   Median :12.98  
##                                                Mean   :3.281   Mean   :12.97  
##                                                3rd Qu.:4.000   3rd Qu.:13.00  
##                                                Max.   :6.000   Max.   :13.10  
##    longitude        Pin code         Output            Feedback        
##  Min.   :77.48   Min.   :560001   Length:388         Length:388        
##  1st Qu.:77.57   1st Qu.:560011   Class :character   Class :character  
##  Median :77.59   Median :560034   Mode  :character   Mode  :character  
##  Mean   :77.60   Mean   :560040                                        
##  3rd Qu.:77.63   3rd Qu.:560068                                        
##  Max.   :77.76   Max.   :560109

Check for na’s and data types.

any(is.na(onlinefood))

## [1] FALSE

sapply(onlinefood, class)

##                        Age                     Gender 
##                  "numeric"                "character" 
##             Marital Status                 Occupation 
##                "character"                "character" 
##             Monthly Income Educational Qualifications 
##                "character"                "character" 
##                Family size                   latitude 
##                  "numeric"                  "numeric" 
##                  longitude                   Pin code 
##                  "numeric"                  "numeric" 
##                     Output                   Feedback 
##                "character"                "character"

Checking Proportions

#checking proportions
table(onlinefood$Occupation)

## 
##       Employee     House wife Self Employeed        Student 
##            118              9             54            207

prop.table(table(onlinefood$Occupation))

## 
##       Employee     House wife Self Employeed        Student 
##     0.30412371     0.02319588     0.13917526     0.53350515

table(onlinefood$`Monthly Income`)

## 
##  10001 to 25000  25001 to 50000  Below Rs.10000 More than 50000       No Income 
##              45              69              25              62             187

prop.table(table(onlinefood$`Monthly Income`))

## 
##  10001 to 25000  25001 to 50000  Below Rs.10000 More than 50000       No Income 
##      0.11597938      0.17783505      0.06443299      0.15979381      0.48195876

table(onlinefood$`Marital Status`)

## 
##           Married Prefer not to say            Single 
##               108                12               268

prop.table(table(onlinefood$`Marital Status`))

## 
##           Married Prefer not to say            Single 
##        0.27835052        0.03092784        0.69072165

table(onlinefood$`Educational Qualifications`)

## 
##      Graduate          Ph.D Post Graduate        School    Uneducated 
##           177            23           174            12             2

prop.table(table(onlinefood$`Educational Qualifications`))

## 
##      Graduate          Ph.D Post Graduate        School    Uneducated 
##   0.456185567   0.059278351   0.448453608   0.030927835   0.005154639

Standard Deviation

#standard deviation
sapply(onlinefood, sd)

##                        Age                     Gender 
##                 2.97559266                         NA 
##             Marital Status                 Occupation 
##                         NA                         NA 
##             Monthly Income Educational Qualifications 
##                         NA                         NA 
##                Family size                   latitude 
##                 1.35102494                 0.04448925 
##                  longitude                   Pin code 
##                 0.05135392                31.39960871 
##                     Output                   Feedback 
##                         NA                         NA

Correlation

cor(onlinefood[c('Age', 'Family size', 'latitude', 'longitude', 'Pin code')])

##                     Age  Family size     latitude   longitude     Pin code
## Age         1.000000000  0.169981512  0.006594549  0.04749956  0.137294125
## Family size 0.169981512  1.000000000 -0.053536733  0.07012644 -0.009402396
## latitude    0.006594549 -0.053536733  1.000000000 -0.14423361 -0.201813497
## longitude   0.047499559  0.070126439 -0.144233613  1.00000000  0.156119479
## Pin code    0.137294125 -0.009402396 -0.201813497  0.15611948  1.000000000

pairs(onlinefood[c('Age', 'Family size', 'latitude', 'longitude', 'Pin code')])

library(psych)
pairs.panels(onlinefood[c('Age', 'Family size', 'latitude', 'longitude', 'Pin code')])

Education Qualification Count

library(ggplot2)

## 
## Attaching package: 'ggplot2'

## The following objects are masked from 'package:psych':
## 
##     %+%, alpha

ggplot(data = onlinefood, aes(x = `Educational Qualifications`, fill = `Educational Qualifications`)) + 
  geom_bar(stat = "count", color = "black") +
  geom_text(stat = "count", aes(label = stat(count)), vjust = -0.5) +
  labs(x = "Educational Qualifications", y = "Count", title = "Educational Qualifications Count") +
  guides(fill = FALSE)

Monthly Income

ggplot(data = onlinefood, aes(x = `Monthly Income`, fill = `Monthly Income`)) + 
  geom_bar(stat = "count", color = "black") +
  geom_text(stat = "count", aes(label = stat(count)), vjust = -0.5) +
  labs(x = "Monthly Income", y = "Count", title = "Monthly Income Count") +
  guides(fill = FALSE)

Age Distribution

ggplot(data = onlinefood, aes(x=Age)) +
  geom_histogram(binwidth = 1, fill = "red", color="black") +
  geom_text(stat = "count", aes(label=stat(count)), vjust = -0.5, color ="black") +
  labs(x = "Age", y = "Frequency", title = "Age Distribution")

Feedback based on output

library(dplyr)

## 
## Attaching package: 'dplyr'

## The following objects are masked from 'package:stats':
## 
##     filter, lag

## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union

library(explore)

## 
## Attaching package: 'explore'

## The following object is masked from 'package:psych':
## 
##     describe

onlinefood %>% explore(Feedback, target=Output)

Positve yes: 94% of the people who said yes had a positive feedback possibly due to timely arrival of their food.
Positive no: 39.1% of the people who said no had a positive feedback possible due to satisfactory resolution to delivery issues or order cancellation.
Negative yes: 6% of the people who said yes had a negative feedback possibly due to complications with food delivery or dislike for the food.
Negative no: 60.9% of the people who said no had a negative feedback possibly due to order cancellation or the order never arriving.

Feedback based on gender

onlinefood %>% explore(Feedback, target=Gender)

The females had slightly more positive feedback(83.7) than the males(80.2%).
The males had more negative feedback(19.8) than the females (16.3).
However there are more males than females in this dataset so regardless of the females having a higher percentage of positive feedbacks, the number of males were still higher than that of females for positive feedback.

Family Size Distribution

ggplot(data = onlinefood, aes(x=`Family size`)) +
  geom_histogram(binwidth = 1, fill = "blue", color="black") +
  geom_text(stat = "count", aes(label=stat(count)), vjust = -0.5, color ="black") +
  labs(x = "Family Size", y = "Frequency", title = "Family Size Distribution")

The majority of family sizes that order food fall between 2 and 4 which possibly indicates a regular couple or a family one parent and one kid or parents with one or two kids…etc

Data Preparation

online_food <- onlinefood
online_food

## # A tibble: 388 × 12
##      Age Gender `Marital Status` Occupation `Monthly Income`
##    <dbl> <chr>  <chr>            <chr>      <chr>           
##  1    20 Female Single           Student    No Income       
##  2    24 Female Single           Student    Below Rs.10000  
##  3    22 Male   Single           Student    Below Rs.10000  
##  4    22 Female Single           Student    No Income       
##  5    22 Male   Single           Student    Below Rs.10000  
##  6    27 Female Married          Employee   More than 50000 
##  7    22 Male   Single           Student    No Income       
##  8    24 Female Single           Student    No Income       
##  9    23 Female Single           Student    No Income       
## 10    23 Female Single           Student    No Income       
## # ℹ 378 more rows
## # ℹ 7 more variables: `Educational Qualifications` <chr>, `Family size` <dbl>,
## #   latitude <dbl>, longitude <dbl>, `Pin code` <dbl>, Output <chr>,
## #   Feedback <chr>

Factors

#gender
str(online_food$Gender)

##  chr [1:388] "Female" "Female" "Male" "Female" "Male" "Female" "Male" ...

online_food$Gender<-factor(online_food$Gender, levels=c("Female","Male"), labels=c("Female", "Male"))
str(online_food$Gender)

##  Factor w/ 2 levels "Female","Male": 1 1 2 1 2 1 2 1 1 1 ...

#marital status
str(online_food$`Marital Status`)

##  chr [1:388] "Single" "Single" "Single" "Single" "Single" "Married" ...

online_food$`Marital Status`<-factor(online_food$`Marital Status`, levels=c("Married", "Prefer not to say","Single"), labels=c("Married", "Prefer not to say","Single") )
str(online_food$`Marital Status`)

##  Factor w/ 3 levels "Married","Prefer not to say",..: 3 3 3 3 3 1 3 3 3 3 ...

#Occupation
str(online_food$Occupation)

##  chr [1:388] "Student" "Student" "Student" "Student" "Student" "Employee" ...

online_food$Occupation<-factor(online_food$Occupation, levels=c("Employee", "House wife", "Self Employeed", "Student"), labels=c("Employee", "House wife", "Self Employeed", "Student") )
str(online_food$Occupation)

##  Factor w/ 4 levels "Employee","House wife",..: 4 4 4 4 4 1 4 4 4 4 ...

#Monthly Income
str(online_food$`Monthly Income`)

##  chr [1:388] "No Income" "Below Rs.10000" "Below Rs.10000" "No Income" ...

online_food$`Monthly Income`<-factor(online_food$`Monthly Income`, levels=c("10001 to 25000", "25001 to 50000", "Below Rs.10000", "More than 50000", "No Income"), labels=c("10001 to 25000", "25001 to 50000", "Below Rs.10000", "More than 50000", "No Income") )
str(online_food$`Monthly Income`)

##  Factor w/ 5 levels "10001 to 25000",..: 5 3 3 5 3 4 5 5 5 5 ...

#Educational Qualifications
str(online_food$`Educational Qualifications`)

##  chr [1:388] "Post Graduate" "Graduate" "Post Graduate" "Graduate" ...

online_food$`Educational Qualifications`<-factor(online_food$`Educational Qualifications`, levels=c("Graduate", "Ph.D", "Post Graduate", "School", "Uneducated"), labels=c("Graduate", "Ph.D", "Post Graduate", "School", "Uneducated"))
str(online_food$`Educational Qualifications`)

##  Factor w/ 5 levels "Graduate","Ph.D",..: 3 1 3 1 3 3 1 3 3 3 ...

#Output
str(online_food$Output)

##  chr [1:388] "Yes" "Yes" "Yes" "Yes" "Yes" "Yes" "Yes" "Yes" "Yes" "Yes" ...

online_food$Output<-factor(online_food$Output, levels=c("No", "Yes"), labels=c("No", "Yes"))
str(online_food$Output)

##  Factor w/ 2 levels "No","Yes": 2 2 2 2 2 2 2 2 2 2 ...

#Feedback
str(online_food$Feedback)

##  chr [1:388] "Positive" "Positive" "Negative" "Positive" "Positive" ...

online_food$Feedback<-factor(online_food$Feedback, levels=c("Negative", "Positive"), labels=c("Negative", "Positive"))
str(online_food$Feedback)

##  Factor w/ 2 levels "Negative","Positive": 2 2 1 2 2 2 2 2 2 2 ...

Check to see if everything matches

str(online_food)

## spc_tbl_ [388 × 12] (S3: spec_tbl_df/tbl_df/tbl/data.frame)
##  $ Age                       : num [1:388] 20 24 22 22 22 27 22 24 23 23 ...
##  $ Gender                    : Factor w/ 2 levels "Female","Male": 1 1 2 1 2 1 2 1 1 1 ...
##  $ Marital Status            : Factor w/ 3 levels "Married","Prefer not to say",..: 3 3 3 3 3 1 3 3 3 3 ...
##  $ Occupation                : Factor w/ 4 levels "Employee","House wife",..: 4 4 4 4 4 1 4 4 4 4 ...
##  $ Monthly Income            : Factor w/ 5 levels "10001 to 25000",..: 5 3 3 5 3 4 5 5 5 5 ...
##  $ Educational Qualifications: Factor w/ 5 levels "Graduate","Ph.D",..: 3 1 3 1 3 3 1 3 3 3 ...
##  $ Family size               : num [1:388] 4 3 3 6 4 2 3 3 2 4 ...
##  $ latitude                  : num [1:388] 13 13 13 12.9 13 ...
##  $ longitude                 : num [1:388] 77.6 77.6 77.7 77.6 77.6 ...
##  $ Pin code                  : num [1:388] 560001 560009 560017 560019 560010 ...
##  $ Output                    : Factor w/ 2 levels "No","Yes": 2 2 2 2 2 2 2 2 2 2 ...
##  $ Feedback                  : Factor w/ 2 levels "Negative","Positive": 2 2 1 2 2 2 2 2 2 2 ...
##  - attr(*, "spec")=
##   .. cols(
##   ..   Age = col_double(),
##   ..   Gender = col_character(),
##   ..   `Marital Status` = col_character(),
##   ..   Occupation = col_character(),
##   ..   `Monthly Income` = col_character(),
##   ..   `Educational Qualifications` = col_character(),
##   ..   `Family size` = col_double(),
##   ..   latitude = col_double(),
##   ..   longitude = col_double(),
##   ..   `Pin code` = col_double(),
##   ..   Output = col_character(),
##   ..   Feedback = col_character()
##   .. )
##  - attr(*, "problems")=<externalptr>

Dependent Variable

online_food$Feedback

##   [1] Positive Positive Negative Positive Positive Positive Positive Positive
##   [9] Positive Positive Positive Negative Positive Positive Positive Positive
##  [17] Positive Negative Negative Positive Positive Positive Positive Positive
##  [25] Positive Positive Positive Positive Positive Positive Positive Positive
##  [33] Positive Positive Positive Positive Positive Negative Positive Positive
##  [41] Negative Positive Positive Positive Positive Positive Positive Positive
##  [49] Negative Positive Positive Positive Positive Positive Positive Positive
##  [57] Positive Positive Positive Positive Positive Positive Positive Positive
##  [65] Positive Positive Positive Positive Positive Positive Positive Positive
##  [73] Positive Positive Positive Positive Positive Positive Positive Positive
##  [81] Positive Positive Positive Positive Positive Positive Positive Positive
##  [89] Positive Negative Positive Positive Positive Positive Positive Positive
##  [97] Positive Positive Positive Positive Positive Positive Positive Positive
## [105] Negative Positive Positive Positive Positive Positive Positive Positive
## [113] Positive Positive Positive Negative Positive Positive Positive Positive
## [121] Positive Positive Positive Negative Negative Positive Positive Positive
## [129] Positive Positive Positive Positive Positive Negative Positive Positive
## [137] Positive Negative Positive Positive Positive Negative Positive Positive
## [145] Negative Positive Positive Positive Positive Positive Positive Positive
## [153] Positive Positive Positive Positive Positive Positive Negative Positive
## [161] Positive Negative Positive Positive Positive Positive Negative Positive
## [169] Positive Negative Positive Positive Positive Positive Positive Positive
## [177] Positive Negative Negative Positive Positive Negative Positive Positive
## [185] Positive Positive Positive Positive Negative Positive Negative Positive
## [193] Positive Positive Positive Positive Positive Positive Positive Negative
## [201] Positive Positive Positive Positive Positive Positive Negative Positive
## [209] Positive Negative Negative Positive Negative Positive Positive Negative
## [217] Positive Negative Positive Positive Positive Positive Positive Positive
## [225] Positive Positive Positive Positive Negative Positive Positive Positive
## [233] Negative Positive Positive Negative Positive Positive Positive Positive
## [241] Positive Positive Positive Positive Negative Positive Negative Positive
## [249] Negative Positive Positive Positive Positive Positive Positive Positive
## [257] Negative Negative Positive Negative Positive Negative Negative Negative
## [265] Positive Positive Negative Positive Positive Positive Positive Negative
## [273] Positive Positive Positive Positive Positive Positive Negative Positive
## [281] Positive Positive Positive Positive Positive Positive Positive Negative
## [289] Positive Positive Positive Positive Negative Negative Negative Positive
## [297] Positive Positive Negative Negative Negative Positive Positive Positive
## [305] Positive Positive Positive Positive Positive Negative Positive Positive
## [313] Positive Positive Positive Negative Positive Positive Positive Negative
## [321] Positive Positive Negative Positive Positive Positive Positive Positive
## [329] Positive Positive Positive Positive Positive Negative Positive Positive
## [337] Negative Positive Positive Negative Positive Positive Negative Positive
## [345] Negative Positive Positive Positive Positive Negative Positive Positive
## [353] Positive Positive Positive Positive Positive Negative Positive Positive
## [361] Positive Negative Positive Positive Negative Positive Positive Positive
## [369] Positive Positive Positive Positive Positive Negative Negative Negative
## [377] Positive Negative Positive Positive Positive Positive Positive Positive
## [385] Positive Positive Positive Positive
## Levels: Negative Positive

table(online_food$Feedback)

## 
## Negative Positive 
##       71      317

prop.table(table(online_food$Feedback))

## 
##  Negative  Positive 
## 0.1829897 0.8170103

Support Vector Machine

library(kernlab)

## 
## Attaching package: 'kernlab'

## The following object is masked from 'package:ggplot2':
## 
##     alpha

## The following object is masked from 'package:psych':
## 
##     alpha

onlinefood_svm<-online_food

Normalization

normalize<-function(x){
  return((x-min(x))/(max(x)-min(x)))
}

normalized_column<-c(1, 7, 8, 9, 10)
onlinefood_svm[normalized_column]<-(lapply(onlinefood_svm[normalized_column], normalize))
summary(onlinefood_svm)

##       Age            Gender              Marital Status          Occupation 
##  Min.   :0.0000   Female:166   Married          :108    Employee      :118  
##  1st Qu.:0.3333   Male  :222   Prefer not to say: 12    House wife    :  9  
##  Median :0.4000                Single           :268    Self Employeed: 54  
##  Mean   :0.4419                                         Student       :207  
##  3rd Qu.:0.5333                                                             
##  Max.   :1.0000                                                             
##          Monthly Income Educational Qualifications  Family size    
##  10001 to 25000 : 45    Graduate     :177          Min.   :0.0000  
##  25001 to 50000 : 69    Ph.D         : 23          1st Qu.:0.2000  
##  Below Rs.10000 : 25    Post Graduate:174          Median :0.4000  
##  More than 50000: 62    School       : 12          Mean   :0.4562  
##  No Income      :187    Uneducated   :  2          3rd Qu.:0.6000  
##                                                    Max.   :1.0000  
##     latitude        longitude         Pin code       Output        Feedback  
##  Min.   :0.0000   Min.   :0.0000   Min.   :0.00000   No : 87   Negative: 71  
##  1st Qu.:0.3028   1st Qu.:0.2959   1st Qu.:0.09028   Yes:301   Positive:317  
##  Median :0.4721   Median :0.3938   Median :0.30093                           
##  Mean   :0.4513   Mean   :0.4232   Mean   :0.36216                           
##  3rd Qu.:0.5567   3rd Qu.:0.5354   3rd Qu.:0.62037                           
##  Max.   :1.0000   Max.   :1.0000   Max.   :1.00000

Create Training and Test Dataset

set.seed(42)
onlinefood_index_svm <- sample(1:nrow(onlinefood_svm), size = floor(0.80*nrow(onlinefood_svm)))
onlinefood_train_svm<-onlinefood_svm[onlinefood_index_svm, ]
onlinefood_test_svm<-onlinefood_svm[-onlinefood_index_svm, ]

Create Classifier

onlinefood_classifier<-ksvm(Feedback~.,data=onlinefood_train_svm, kernel="vanilladot")

##  Setting default kernel parameters

onlinefood_classifier

## Support Vector Machine object of class "ksvm" 
## 
## SV type: C-svc  (classification) 
##  parameter : cost C = 1 
## 
## Linear (vanilla) kernel function. 
## 
## Number of Support Vectors : 137 
## 
## Objective Function Value : -82 
## Training error : 0.129032

Evaluate Onlinefood Prediction

onlinefood_predictions<-predict(onlinefood_classifier, onlinefood_test_svm)
onlinefood_predictions

##  [1] Positive Positive Positive Positive Positive Positive Positive Positive
##  [9] Positive Positive Positive Positive Positive Positive Positive Positive
## [17] Positive Positive Positive Positive Negative Negative Positive Positive
## [25] Positive Positive Positive Negative Positive Negative Negative Positive
## [33] Positive Positive Positive Positive Positive Positive Positive Positive
## [41] Negative Positive Negative Positive Negative Negative Negative Positive
## [49] Positive Positive Negative Positive Positive Positive Positive Positive
## [57] Negative Negative Negative Positive Positive Negative Negative Negative
## [65] Positive Positive Positive Negative Positive Positive Positive Negative
## [73] Positive Positive Positive Negative Negative Positive
## Levels: Negative Positive

Understanding Classification Results

table(onlinefood_predictions, onlinefood_test_svm$Feedback)

##                       
## onlinefood_predictions Negative Positive
##               Negative       13        8
##               Positive        4       53

o_food<-onlinefood_predictions==onlinefood_test_svm$Feedback
o_food

##  [1]  TRUE  TRUE  TRUE FALSE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE
## [13]  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE FALSE FALSE  TRUE  TRUE
## [25]  TRUE  TRUE  TRUE  TRUE  TRUE FALSE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE
## [37]  TRUE  TRUE  TRUE  TRUE FALSE FALSE FALSE  TRUE  TRUE  TRUE  TRUE FALSE
## [49]  TRUE  TRUE FALSE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE FALSE  TRUE  TRUE
## [61]  TRUE  TRUE  TRUE  TRUE FALSE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE
## [73]  TRUE  TRUE  TRUE FALSE  TRUE  TRUE

table(o_food)

## o_food
## FALSE  TRUE 
##    12    66

There were 12 incorrectly classified and 66 correctly classified feedbacks.

prop.table(table(o_food))

## o_food
##     FALSE      TRUE 
## 0.1538462 0.8461538

#15% incorrectly classified and 84% correctly classified

15% of the feedbacks were incorrectly classified and 84% of the feedbacks were correctly classified.

Improving model performance

The previous SVM model used the simple linear kernel function.
Using a more complex kernel function, we can map the data into a higher dimensional space and potentially obtain a better model fit.
Using Gaussian RBF kernel which has shown to perform well for many types of data.

onlinefood_classifier_rbf<-ksvm(Feedback~.,data=onlinefood_train_svm, kernel="rbfdot")
onlinefood_classifier_rbf

## Support Vector Machine object of class "ksvm" 
## 
## SV type: C-svc  (classification) 
##  parameter : cost C = 1 
## 
## Gaussian Radial Basis kernel function. 
##  Hyperparameter : sigma =  0.0589947205070788 
## 
## Number of Support Vectors : 138 
## 
## Objective Function Value : -88.8184 
## Training error : 0.122581

Prediction

onlinefood_predictions_rbf<-predict(onlinefood_classifier_rbf,onlinefood_test_svm)
onlinefood_predictions_rbf

##  [1] Positive Positive Positive Positive Positive Positive Positive Positive
##  [9] Positive Positive Positive Positive Positive Positive Positive Positive
## [17] Positive Positive Positive Positive Positive Positive Positive Positive
## [25] Positive Positive Positive Positive Positive Positive Positive Positive
## [33] Positive Positive Positive Positive Positive Positive Positive Positive
## [41] Positive Positive Positive Positive Positive Negative Positive Positive
## [49] Positive Positive Positive Positive Positive Positive Positive Positive
## [57] Negative Positive Positive Positive Positive Positive Positive Positive
## [65] Positive Positive Positive Positive Positive Positive Positive Positive
## [73] Positive Positive Positive Positive Negative Positive
## Levels: Negative Positive

Classification Results

table(onlinefood_predictions_rbf, onlinefood_test_svm$Feedback)

##                           
## onlinefood_predictions_rbf Negative Positive
##                   Negative        3        0
##                   Positive       14       61

o_food_rbf<-onlinefood_predictions_rbf==onlinefood_test_svm$Feedback
o_food_rbf

##  [1]  TRUE  TRUE  TRUE FALSE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE
## [13]  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE
## [25]  TRUE  TRUE  TRUE FALSE  TRUE  TRUE FALSE  TRUE  TRUE  TRUE  TRUE  TRUE
## [37]  TRUE  TRUE  TRUE  TRUE  TRUE FALSE  TRUE  TRUE FALSE  TRUE FALSE FALSE
## [49]  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE FALSE  TRUE
## [61]  TRUE FALSE FALSE FALSE FALSE  TRUE  TRUE FALSE  TRUE  TRUE  TRUE FALSE
## [73]  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE

table(o_food_rbf)

## o_food_rbf
## FALSE  TRUE 
##    14    64

There were 14 incorrectly classified feedbacks and 64 correctly classified feedbacks.
The linear kernel function performed better.

prop.table(table(o_food_rbf))

## o_food_rbf
##     FALSE      TRUE 
## 0.1794872 0.8205128

There were 17% incorrectly classified and 82% correctly classified.

Confusion Matrix

library(gmodels)
CrossTable(x = onlinefood_test_svm$Feedback, y=onlinefood_predictions, prop.chisq=FALSE)

## 
##  
##    Cell Contents
## |-------------------------|
## |                       N |
## |           N / Row Total |
## |           N / Col Total |
## |         N / Table Total |
## |-------------------------|
## 
##  
## Total Observations in Table:  78 
## 
##  
##                              | onlinefood_predictions 
## onlinefood_test_svm$Feedback |  Negative |  Positive | Row Total | 
## -----------------------------|-----------|-----------|-----------|
##                     Negative |        13 |         4 |        17 | 
##                              |     0.765 |     0.235 |     0.218 | 
##                              |     0.619 |     0.070 |           | 
##                              |     0.167 |     0.051 |           | 
## -----------------------------|-----------|-----------|-----------|
##                     Positive |         8 |        53 |        61 | 
##                              |     0.131 |     0.869 |     0.782 | 
##                              |     0.381 |     0.930 |           | 
##                              |     0.103 |     0.679 |           | 
## -----------------------------|-----------|-----------|-----------|
##                 Column Total |        21 |        57 |        78 | 
##                              |     0.269 |     0.731 |           | 
## -----------------------------|-----------|-----------|-----------|
## 
##

KNN

onlinefood_knn<-onlinefood
onlinefood_knn

## # A tibble: 388 × 12
##      Age Gender `Marital Status` Occupation `Monthly Income`
##    <dbl> <chr>  <chr>            <chr>      <chr>           
##  1    20 Female Single           Student    No Income       
##  2    24 Female Single           Student    Below Rs.10000  
##  3    22 Male   Single           Student    Below Rs.10000  
##  4    22 Female Single           Student    No Income       
##  5    22 Male   Single           Student    Below Rs.10000  
##  6    27 Female Married          Employee   More than 50000 
##  7    22 Male   Single           Student    No Income       
##  8    24 Female Single           Student    No Income       
##  9    23 Female Single           Student    No Income       
## 10    23 Female Single           Student    No Income       
## # ℹ 378 more rows
## # ℹ 7 more variables: `Educational Qualifications` <chr>, `Family size` <dbl>,
## #   latitude <dbl>, longitude <dbl>, `Pin code` <dbl>, Output <chr>,
## #   Feedback <chr>

Data Preparation

#tweak the factors for the factor columns
#Gender
str(onlinefood_knn$Gender)

##  chr [1:388] "Female" "Female" "Male" "Female" "Male" "Female" "Male" ...

onlinefood_knn$Gender<-factor(onlinefood_knn$Gender, levels=c("Female","Male"), labels=c("1", "2"))
#1 for female, 2 for male
str(onlinefood_knn$Gender)

##  Factor w/ 2 levels "1","2": 1 1 2 1 2 1 2 1 1 1 ...

#change from factor to numeric
onlinefood_knn$Gender<-as.numeric(onlinefood_knn$Gender)
#confirm
str(onlinefood_knn$Gender)

##  num [1:388] 1 1 2 1 2 1 2 1 1 1 ...

#Marital Status
str(onlinefood_knn$`Marital Status`)

##  chr [1:388] "Single" "Single" "Single" "Single" "Single" "Married" ...

onlinefood_knn$`Marital Status`<-factor(onlinefood_knn$`Marital Status`, levels=c("Married", "Prefer not to say","Single"), labels=c("1", "2","3") )
str(onlinefood_knn$`Marital Status`)

##  Factor w/ 3 levels "1","2","3": 3 3 3 3 3 1 3 3 3 3 ...

#factor to numeric
onlinefood_knn$`Marital Status`<-as.numeric(onlinefood_knn$`Marital Status`)
#confirm
str(onlinefood_knn$`Marital Status`)

##  num [1:388] 3 3 3 3 3 1 3 3 3 3 ...

#Occupation
str(onlinefood_knn$Occupation)

##  chr [1:388] "Student" "Student" "Student" "Student" "Student" "Employee" ...

onlinefood_knn$Occupation<-factor(onlinefood_knn$Occupation, levels=c("Employee", "House wife", "Self Employeed", "Student"), labels=c("1", "2", "3", "4") )
str(onlinefood_knn$Occupation)

##  Factor w/ 4 levels "1","2","3","4": 4 4 4 4 4 1 4 4 4 4 ...

#factor to numeric
onlinefood_knn$Occupation<-as.numeric(onlinefood_knn$Occupation)
#confirm
str(onlinefood_knn$Occupation)

##  num [1:388] 4 4 4 4 4 1 4 4 4 4 ...

#Monthly Income
str(onlinefood_knn$`Monthly Income`)

##  chr [1:388] "No Income" "Below Rs.10000" "Below Rs.10000" "No Income" ...

onlinefood_knn$`Monthly Income`<-factor(onlinefood_knn$`Monthly Income`, levels=c("10001 to 25000", "25001 to 50000", "Below Rs.10000", "More than 50000", "No Income"), labels=c("1", "2", "3", "4", "5") )
str(onlinefood_knn$`Monthly Income`)

##  Factor w/ 5 levels "1","2","3","4",..: 5 3 3 5 3 4 5 5 5 5 ...

#factor to numeric
onlinefood_knn$`Monthly Income`<-as.numeric(onlinefood_knn$`Monthly Income`)
#confirm
str(onlinefood_knn$`Monthly Income`)

##  num [1:388] 5 3 3 5 3 4 5 5 5 5 ...

#Educational Qualifications
str(onlinefood_knn$`Educational Qualifications`)

##  chr [1:388] "Post Graduate" "Graduate" "Post Graduate" "Graduate" ...

onlinefood_knn$`Educational Qualifications`<-factor(onlinefood_knn$`Educational Qualifications`, levels=c("Graduate", "Ph.D", "Post Graduate", "School", "Uneducated"), labels=c("1", "2", "3", "4", "5"))
str(onlinefood_knn$`Educational Qualifications`)

##  Factor w/ 5 levels "1","2","3","4",..: 3 1 3 1 3 3 1 3 3 3 ...

#factor to numeric
onlinefood_knn$`Educational Qualifications`<-as.numeric(onlinefood_knn$`Educational Qualifications`)
#confirm
str(onlinefood_knn$`Educational Qualifications`)

##  num [1:388] 3 1 3 1 3 3 1 3 3 3 ...

#Output
str(onlinefood_knn$Output)

##  chr [1:388] "Yes" "Yes" "Yes" "Yes" "Yes" "Yes" "Yes" "Yes" "Yes" "Yes" ...

onlinefood_knn$Output<-factor(onlinefood_knn$Output, levels=c("No", "Yes"), labels=c("1", "2"))
str(onlinefood_knn$Output)

##  Factor w/ 2 levels "1","2": 2 2 2 2 2 2 2 2 2 2 ...

#factor to numeric
onlinefood_knn$Output<-as.numeric(onlinefood_knn$Output)
#confirm
str(onlinefood_knn$Output)

##  num [1:388] 2 2 2 2 2 2 2 2 2 2 ...

#Feedback
str(onlinefood_knn$Feedback)

##  chr [1:388] "Positive" "Positive" "Negative" "Positive" "Positive" ...

onlinefood_knn$Feedback<-factor(onlinefood_knn$Feedback, levels=c("Negative", "Positive"), labels=c("Negative", "Positive"))
str(onlinefood_knn$Feedback)

##  Factor w/ 2 levels "Negative","Positive": 2 2 1 2 2 2 2 2 2 2 ...

Check structure of dataset

str(onlinefood_knn)

## spc_tbl_ [388 × 12] (S3: spec_tbl_df/tbl_df/tbl/data.frame)
##  $ Age                       : num [1:388] 20 24 22 22 22 27 22 24 23 23 ...
##  $ Gender                    : num [1:388] 1 1 2 1 2 1 2 1 1 1 ...
##  $ Marital Status            : num [1:388] 3 3 3 3 3 1 3 3 3 3 ...
##  $ Occupation                : num [1:388] 4 4 4 4 4 1 4 4 4 4 ...
##  $ Monthly Income            : num [1:388] 5 3 3 5 3 4 5 5 5 5 ...
##  $ Educational Qualifications: num [1:388] 3 1 3 1 3 3 1 3 3 3 ...
##  $ Family size               : num [1:388] 4 3 3 6 4 2 3 3 2 4 ...
##  $ latitude                  : num [1:388] 13 13 13 12.9 13 ...
##  $ longitude                 : num [1:388] 77.6 77.6 77.7 77.6 77.6 ...
##  $ Pin code                  : num [1:388] 560001 560009 560017 560019 560010 ...
##  $ Output                    : num [1:388] 2 2 2 2 2 2 2 2 2 2 ...
##  $ Feedback                  : Factor w/ 2 levels "Negative","Positive": 2 2 1 2 2 2 2 2 2 2 ...
##  - attr(*, "spec")=
##   .. cols(
##   ..   Age = col_double(),
##   ..   Gender = col_character(),
##   ..   `Marital Status` = col_character(),
##   ..   Occupation = col_character(),
##   ..   `Monthly Income` = col_character(),
##   ..   `Educational Qualifications` = col_character(),
##   ..   `Family size` = col_double(),
##   ..   latitude = col_double(),
##   ..   longitude = col_double(),
##   ..   `Pin code` = col_double(),
##   ..   Output = col_character(),
##   ..   Feedback = col_character()
##   .. )
##  - attr(*, "problems")=<externalptr>

Normalization

normalize<-function(x){
  return((x-min(x))/(max(x)-min(x)))
}

Training and Test Dataset

#before that, create a new dataset without the labels and one with labels
summary(onlinefood_knn)

##       Age            Gender      Marital Status    Occupation   
##  Min.   :18.00   Min.   :1.000   Min.   :1.000   Min.   :1.000  
##  1st Qu.:23.00   1st Qu.:1.000   1st Qu.:1.000   1st Qu.:1.000  
##  Median :24.00   Median :2.000   Median :3.000   Median :4.000  
##  Mean   :24.63   Mean   :1.572   Mean   :2.412   Mean   :2.902  
##  3rd Qu.:26.00   3rd Qu.:2.000   3rd Qu.:3.000   3rd Qu.:4.000  
##  Max.   :33.00   Max.   :2.000   Max.   :3.000   Max.   :4.000  
##  Monthly Income  Educational Qualifications  Family size       latitude    
##  Min.   :1.000   Min.   :1.00               Min.   :1.000   Min.   :12.87  
##  1st Qu.:2.000   1st Qu.:1.00               1st Qu.:2.000   1st Qu.:12.94  
##  Median :4.000   Median :2.00               Median :3.000   Median :12.98  
##  Mean   :3.714   Mean   :2.07               Mean   :3.281   Mean   :12.97  
##  3rd Qu.:5.000   3rd Qu.:3.00               3rd Qu.:4.000   3rd Qu.:13.00  
##  Max.   :5.000   Max.   :5.00               Max.   :6.000   Max.   :13.10  
##    longitude        Pin code          Output          Feedback  
##  Min.   :77.48   Min.   :560001   Min.   :1.000   Negative: 71  
##  1st Qu.:77.57   1st Qu.:560011   1st Qu.:2.000   Positive:317  
##  Median :77.59   Median :560034   Median :2.000                 
##  Mean   :77.60   Mean   :560040   Mean   :1.776                 
##  3rd Qu.:77.63   3rd Qu.:560068   3rd Qu.:2.000                 
##  Max.   :77.76   Max.   :560109   Max.   :2.000

online_food1<-onlinefood_knn[-12]
online_food1#dataset without dependent variable

## # A tibble: 388 × 11
##      Age Gender `Marital Status` Occupation `Monthly Income`
##    <dbl>  <dbl>            <dbl>      <dbl>            <dbl>
##  1    20      1                3          4                5
##  2    24      1                3          4                3
##  3    22      2                3          4                3
##  4    22      1                3          4                5
##  5    22      2                3          4                3
##  6    27      1                1          1                4
##  7    22      2                3          4                5
##  8    24      1                3          4                5
##  9    23      1                3          4                5
## 10    23      1                3          4                5
## # ℹ 378 more rows
## # ℹ 6 more variables: `Educational Qualifications` <dbl>, `Family size` <dbl>,
## #   latitude <dbl>, longitude <dbl>, `Pin code` <dbl>, Output <dbl>

summary(online_food1)

##       Age            Gender      Marital Status    Occupation   
##  Min.   :18.00   Min.   :1.000   Min.   :1.000   Min.   :1.000  
##  1st Qu.:23.00   1st Qu.:1.000   1st Qu.:1.000   1st Qu.:1.000  
##  Median :24.00   Median :2.000   Median :3.000   Median :4.000  
##  Mean   :24.63   Mean   :1.572   Mean   :2.412   Mean   :2.902  
##  3rd Qu.:26.00   3rd Qu.:2.000   3rd Qu.:3.000   3rd Qu.:4.000  
##  Max.   :33.00   Max.   :2.000   Max.   :3.000   Max.   :4.000  
##  Monthly Income  Educational Qualifications  Family size       latitude    
##  Min.   :1.000   Min.   :1.00               Min.   :1.000   Min.   :12.87  
##  1st Qu.:2.000   1st Qu.:1.00               1st Qu.:2.000   1st Qu.:12.94  
##  Median :4.000   Median :2.00               Median :3.000   Median :12.98  
##  Mean   :3.714   Mean   :2.07               Mean   :3.281   Mean   :12.97  
##  3rd Qu.:5.000   3rd Qu.:3.00               3rd Qu.:4.000   3rd Qu.:13.00  
##  Max.   :5.000   Max.   :5.00               Max.   :6.000   Max.   :13.10  
##    longitude        Pin code          Output     
##  Min.   :77.48   Min.   :560001   Min.   :1.000  
##  1st Qu.:77.57   1st Qu.:560011   1st Qu.:2.000  
##  Median :77.59   Median :560034   Median :2.000  
##  Mean   :77.60   Mean   :560040   Mean   :1.776  
##  3rd Qu.:77.63   3rd Qu.:560068   3rd Qu.:2.000  
##  Max.   :77.76   Max.   :560109   Max.   :2.000

Standardize the dataset without labels

online_food1<-as.data.frame(lapply(online_food1, normalize))
summary(online_food1)

##       Age             Gender       Marital.Status     Occupation   
##  Min.   :0.0000   Min.   :0.0000   Min.   :0.0000   Min.   :0.000  
##  1st Qu.:0.3333   1st Qu.:0.0000   1st Qu.:0.0000   1st Qu.:0.000  
##  Median :0.4000   Median :1.0000   Median :1.0000   Median :1.000  
##  Mean   :0.4419   Mean   :0.5722   Mean   :0.7062   Mean   :0.634  
##  3rd Qu.:0.5333   3rd Qu.:1.0000   3rd Qu.:1.0000   3rd Qu.:1.000  
##  Max.   :1.0000   Max.   :1.0000   Max.   :1.0000   Max.   :1.000  
##  Monthly.Income   Educational.Qualifications  Family.size        latitude     
##  Min.   :0.0000   Min.   :0.0000             Min.   :0.0000   Min.   :0.0000  
##  1st Qu.:0.2500   1st Qu.:0.0000             1st Qu.:0.2000   1st Qu.:0.3028  
##  Median :0.7500   Median :0.2500             Median :0.4000   Median :0.4721  
##  Mean   :0.6785   Mean   :0.2674             Mean   :0.4562   Mean   :0.4513  
##  3rd Qu.:1.0000   3rd Qu.:0.5000             3rd Qu.:0.6000   3rd Qu.:0.5567  
##  Max.   :1.0000   Max.   :1.0000             Max.   :1.0000   Max.   :1.0000  
##    longitude         Pin.code           Output      
##  Min.   :0.0000   Min.   :0.00000   Min.   :0.0000  
##  1st Qu.:0.2959   1st Qu.:0.09028   1st Qu.:1.0000  
##  Median :0.3938   Median :0.30093   Median :1.0000  
##  Mean   :0.4232   Mean   :0.36216   Mean   :0.7758  
##  3rd Qu.:0.5354   3rd Qu.:0.62037   3rd Qu.:1.0000  
##  Max.   :1.0000   Max.   :1.00000   Max.   :1.0000

Labelled Dataset

onlinefood_labels<-online_food
onlinefood_labels

## # A tibble: 388 × 12
##      Age Gender `Marital Status` Occupation `Monthly Income`
##    <dbl> <fct>  <fct>            <fct>      <fct>           
##  1    20 Female Single           Student    No Income       
##  2    24 Female Single           Student    Below Rs.10000  
##  3    22 Male   Single           Student    Below Rs.10000  
##  4    22 Female Single           Student    No Income       
##  5    22 Male   Single           Student    Below Rs.10000  
##  6    27 Female Married          Employee   More than 50000 
##  7    22 Male   Single           Student    No Income       
##  8    24 Female Single           Student    No Income       
##  9    23 Female Single           Student    No Income       
## 10    23 Female Single           Student    No Income       
## # ℹ 378 more rows
## # ℹ 7 more variables: `Educational Qualifications` <fct>, `Family size` <dbl>,
## #   latitude <dbl>, longitude <dbl>, `Pin code` <dbl>, Output <fct>,
## #   Feedback <fct>

Training & Test (No label)

onlinefood_train<-online_food1[1:310, ]
onlinefood_test<-online_food1[311:388, ]

Training & Test (Labelled)

onlinefood_train_labels<-onlinefood_labels[1:310, 12]
onlinefood_test_labels<-onlinefood_labels[311:388, 12]

Choosing best k-value

library(class)
predictions <- list()

# Loop through each k from 2 to 18
for (k in 2:18) {
  # Run knn with the current value of k
  predictions[[paste("k=", k, sep="")]] <- knn(train=onlinefood_train, test=onlinefood_test, cl=onlinefood_train_labels$Feedback, k=k)
}
predictions

## $`k=2`
##  [1] Positive Positive Positive Positive Positive Negative Positive Positive
##  [9] Positive Positive Positive Positive Positive Positive Positive Positive
## [17] Positive Positive Positive Positive Positive Positive Positive Positive
## [25] Positive Positive Negative Negative Positive Positive Positive Positive
## [33] Positive Positive Negative Positive Positive Positive Positive Positive
## [41] Positive Positive Negative Negative Positive Negative Positive Negative
## [49] Positive Positive Positive Positive Positive Positive Positive Positive
## [57] Positive Positive Positive Positive Positive Positive Positive Negative
## [65] Positive Negative Positive Negative Positive Positive Positive Positive
## [73] Positive Positive Positive Positive Negative Positive
## Levels: Negative Positive
## 
## $`k=3`
##  [1] Positive Positive Positive Positive Positive Negative Positive Positive
##  [9] Positive Positive Positive Positive Negative Positive Positive Positive
## [17] Positive Positive Positive Positive Positive Positive Positive Negative
## [25] Positive Positive Negative Negative Positive Negative Positive Positive
## [33] Positive Positive Negative Positive Positive Positive Positive Negative
## [41] Positive Positive Negative Negative Positive Negative Positive Negative
## [49] Positive Positive Positive Positive Positive Positive Negative Positive
## [57] Positive Positive Positive Positive Positive Positive Positive Negative
## [65] Positive Negative Positive Negative Positive Positive Positive Positive
## [73] Positive Positive Positive Positive Negative Positive
## Levels: Negative Positive
## 
## $`k=4`
##  [1] Positive Positive Positive Positive Positive Negative Positive Positive
##  [9] Positive Positive Positive Positive Negative Positive Positive Positive
## [17] Positive Positive Positive Positive Positive Positive Positive Negative
## [25] Positive Positive Negative Negative Positive Negative Positive Positive
## [33] Positive Positive Negative Positive Positive Positive Positive Positive
## [41] Positive Positive Negative Negative Positive Positive Positive Negative
## [49] Positive Positive Positive Positive Positive Positive Negative Positive
## [57] Positive Positive Negative Positive Positive Positive Positive Negative
## [65] Positive Negative Positive Negative Positive Positive Positive Positive
## [73] Positive Positive Positive Positive Positive Positive
## Levels: Negative Positive
## 
## $`k=5`
##  [1] Positive Positive Positive Positive Positive Negative Positive Positive
##  [9] Positive Positive Positive Positive Negative Positive Positive Positive
## [17] Positive Positive Positive Positive Positive Positive Positive Negative
## [25] Positive Positive Negative Negative Positive Negative Positive Positive
## [33] Positive Positive Negative Positive Positive Positive Positive Negative
## [41] Positive Positive Negative Negative Positive Negative Positive Negative
## [49] Positive Positive Positive Positive Positive Positive Negative Positive
## [57] Positive Positive Positive Positive Positive Positive Positive Negative
## [65] Positive Negative Positive Negative Positive Positive Positive Positive
## [73] Positive Positive Positive Positive Positive Positive
## Levels: Negative Positive
## 
## $`k=6`
##  [1] Positive Positive Positive Positive Positive Negative Positive Positive
##  [9] Positive Positive Positive Positive Negative Positive Positive Positive
## [17] Positive Positive Positive Positive Positive Positive Positive Negative
## [25] Positive Positive Negative Negative Positive Negative Positive Positive
## [33] Positive Positive Negative Positive Positive Positive Positive Negative
## [41] Positive Positive Negative Negative Positive Negative Positive Negative
## [49] Positive Positive Positive Positive Positive Positive Negative Positive
## [57] Positive Positive Positive Positive Positive Positive Positive Negative
## [65] Positive Negative Positive Negative Positive Positive Positive Positive
## [73] Positive Positive Positive Positive Negative Positive
## Levels: Negative Positive
## 
## $`k=7`
##  [1] Positive Positive Positive Positive Positive Negative Positive Positive
##  [9] Positive Negative Positive Positive Negative Positive Positive Positive
## [17] Positive Positive Positive Positive Positive Positive Positive Negative
## [25] Positive Positive Negative Negative Positive Negative Positive Positive
## [33] Positive Positive Negative Positive Positive Positive Positive Negative
## [41] Positive Positive Negative Negative Positive Negative Positive Negative
## [49] Positive Positive Positive Negative Positive Positive Negative Positive
## [57] Positive Positive Positive Positive Positive Positive Positive Negative
## [65] Positive Negative Positive Negative Positive Positive Positive Positive
## [73] Positive Positive Positive Positive Negative Positive
## Levels: Negative Positive
## 
## $`k=8`
##  [1] Positive Positive Positive Positive Positive Negative Positive Positive
##  [9] Positive Negative Positive Positive Negative Positive Positive Positive
## [17] Positive Positive Positive Positive Positive Positive Positive Negative
## [25] Positive Positive Negative Negative Positive Negative Positive Positive
## [33] Positive Positive Negative Positive Positive Positive Positive Negative
## [41] Positive Positive Negative Negative Positive Negative Positive Negative
## [49] Positive Positive Positive Positive Positive Positive Negative Positive
## [57] Positive Positive Positive Positive Positive Positive Positive Negative
## [65] Positive Negative Positive Negative Positive Positive Positive Positive
## [73] Positive Positive Positive Positive Positive Positive
## Levels: Negative Positive
## 
## $`k=9`
##  [1] Positive Positive Positive Positive Positive Negative Positive Positive
##  [9] Positive Negative Positive Positive Negative Positive Positive Positive
## [17] Positive Positive Positive Positive Positive Positive Positive Negative
## [25] Positive Positive Negative Negative Positive Negative Positive Positive
## [33] Positive Positive Negative Positive Positive Positive Positive Negative
## [41] Positive Positive Negative Negative Positive Negative Positive Negative
## [49] Positive Positive Positive Negative Positive Positive Negative Positive
## [57] Positive Positive Positive Positive Positive Positive Positive Negative
## [65] Positive Negative Positive Negative Positive Positive Positive Positive
## [73] Positive Positive Positive Positive Positive Positive
## Levels: Negative Positive
## 
## $`k=10`
##  [1] Positive Positive Positive Positive Positive Negative Positive Positive
##  [9] Positive Negative Positive Positive Negative Positive Positive Positive
## [17] Positive Positive Positive Positive Positive Positive Positive Negative
## [25] Positive Positive Negative Negative Positive Negative Positive Positive
## [33] Positive Positive Negative Positive Positive Positive Positive Negative
## [41] Positive Positive Negative Negative Positive Negative Positive Negative
## [49] Positive Positive Positive Negative Positive Positive Negative Positive
## [57] Positive Positive Positive Positive Positive Positive Positive Negative
## [65] Positive Negative Positive Negative Positive Positive Positive Positive
## [73] Positive Positive Positive Positive Positive Positive
## Levels: Negative Positive
## 
## $`k=11`
##  [1] Positive Positive Positive Positive Positive Negative Positive Positive
##  [9] Positive Positive Positive Positive Negative Positive Positive Positive
## [17] Positive Positive Positive Positive Positive Positive Positive Negative
## [25] Positive Positive Negative Negative Positive Negative Positive Positive
## [33] Positive Positive Negative Positive Positive Positive Positive Negative
## [41] Positive Positive Negative Negative Positive Negative Positive Negative
## [49] Positive Positive Positive Positive Positive Positive Negative Positive
## [57] Positive Positive Positive Positive Positive Positive Positive Negative
## [65] Positive Negative Positive Negative Positive Positive Positive Positive
## [73] Positive Positive Positive Positive Positive Positive
## Levels: Negative Positive
## 
## $`k=12`
##  [1] Positive Positive Positive Positive Positive Negative Positive Positive
##  [9] Positive Positive Positive Positive Negative Positive Positive Positive
## [17] Positive Positive Positive Positive Positive Positive Positive Negative
## [25] Positive Positive Negative Negative Positive Negative Positive Positive
## [33] Positive Positive Negative Positive Positive Positive Positive Negative
## [41] Positive Positive Negative Negative Positive Negative Positive Negative
## [49] Positive Positive Positive Negative Positive Positive Negative Positive
## [57] Positive Positive Positive Positive Positive Positive Positive Negative
## [65] Positive Positive Positive Negative Positive Positive Positive Positive
## [73] Positive Positive Positive Positive Positive Positive
## Levels: Negative Positive
## 
## $`k=13`
##  [1] Positive Positive Positive Positive Positive Negative Positive Positive
##  [9] Positive Negative Positive Positive Negative Positive Positive Positive
## [17] Positive Positive Positive Positive Positive Positive Positive Negative
## [25] Positive Positive Negative Negative Positive Negative Positive Positive
## [33] Positive Positive Negative Positive Positive Positive Positive Negative
## [41] Positive Positive Negative Negative Positive Negative Positive Negative
## [49] Positive Positive Positive Negative Positive Positive Negative Positive
## [57] Positive Positive Positive Positive Positive Positive Positive Negative
## [65] Positive Positive Positive Negative Positive Positive Positive Positive
## [73] Positive Positive Positive Positive Positive Positive
## Levels: Negative Positive
## 
## $`k=14`
##  [1] Positive Positive Positive Positive Positive Negative Positive Positive
##  [9] Positive Negative Positive Positive Negative Positive Positive Positive
## [17] Positive Positive Positive Positive Positive Positive Positive Negative
## [25] Positive Positive Negative Negative Positive Negative Positive Positive
## [33] Positive Positive Positive Positive Positive Positive Positive Negative
## [41] Positive Positive Negative Negative Positive Negative Positive Negative
## [49] Positive Positive Positive Positive Positive Positive Negative Positive
## [57] Positive Positive Positive Positive Positive Positive Positive Negative
## [65] Positive Positive Positive Negative Positive Positive Positive Positive
## [73] Positive Positive Positive Positive Positive Positive
## Levels: Negative Positive
## 
## $`k=15`
##  [1] Positive Positive Positive Positive Positive Negative Positive Positive
##  [9] Positive Negative Positive Positive Negative Positive Positive Positive
## [17] Positive Positive Positive Positive Positive Positive Positive Negative
## [25] Positive Positive Negative Negative Positive Negative Positive Positive
## [33] Positive Positive Positive Positive Positive Positive Positive Negative
## [41] Positive Positive Negative Negative Positive Negative Positive Negative
## [49] Positive Positive Positive Negative Positive Positive Negative Positive
## [57] Positive Positive Positive Positive Positive Positive Positive Negative
## [65] Positive Positive Positive Negative Positive Positive Positive Positive
## [73] Positive Positive Positive Positive Positive Positive
## Levels: Negative Positive
## 
## $`k=16`
##  [1] Positive Positive Positive Positive Positive Positive Positive Positive
##  [9] Positive Negative Positive Positive Negative Positive Positive Positive
## [17] Positive Positive Positive Positive Positive Positive Positive Negative
## [25] Positive Positive Negative Negative Positive Negative Positive Positive
## [33] Positive Positive Negative Positive Positive Positive Positive Negative
## [41] Positive Positive Negative Negative Positive Negative Positive Negative
## [49] Positive Positive Positive Positive Positive Positive Negative Positive
## [57] Positive Positive Positive Positive Positive Positive Positive Negative
## [65] Positive Positive Positive Negative Positive Positive Positive Positive
## [73] Positive Positive Positive Positive Positive Positive
## Levels: Negative Positive
## 
## $`k=17`
##  [1] Positive Positive Positive Positive Positive Negative Positive Positive
##  [9] Positive Positive Positive Positive Negative Positive Positive Positive
## [17] Positive Positive Positive Positive Positive Positive Positive Negative
## [25] Positive Positive Negative Negative Positive Positive Positive Positive
## [33] Positive Positive Positive Positive Positive Positive Positive Negative
## [41] Positive Positive Negative Negative Positive Positive Positive Negative
## [49] Positive Positive Positive Positive Positive Positive Negative Positive
## [57] Positive Positive Positive Positive Positive Positive Positive Negative
## [65] Positive Negative Positive Negative Positive Positive Positive Positive
## [73] Positive Positive Positive Positive Positive Positive
## Levels: Negative Positive
## 
## $`k=18`
##  [1] Positive Positive Positive Positive Positive Positive Positive Positive
##  [9] Positive Positive Positive Positive Negative Positive Positive Positive
## [17] Positive Positive Positive Positive Positive Positive Positive Negative
## [25] Positive Positive Negative Negative Positive Positive Positive Positive
## [33] Positive Positive Positive Positive Positive Positive Positive Negative
## [41] Positive Positive Negative Negative Positive Positive Positive Negative
## [49] Positive Positive Positive Positive Positive Positive Negative Positive
## [57] Positive Positive Positive Positive Positive Positive Positive Negative
## [65] Positive Positive Positive Negative Positive Positive Positive Positive
## [73] Positive Positive Positive Positive Positive Positive
## Levels: Negative Positive

library(dplyr)
accuracies <- sapply(predictions, function(pred, true_labels) {
  mean(pred == true_labels)
}, true_labels = onlinefood_test_labels$Feedback)

accuracy_data <- data.frame(
  k = 2:18,
  Accuracy = accuracies
)

Visualizing k-values

library(ggplot2)
ggplot(accuracy_data, aes(x = k, y = Accuracy)) +
  geom_line() + # Line plot to show the trend
  geom_point() + # Points to mark each k-value's accuracy
  labs(title = "Accuracy vs. k in k-NN", x = "k (Number of Neighbors)", y = "Accuracy") +
  theme_minimal()

The best value for k is 9 as it has the highest accuracy out of all the values between 2-18.

Evaluate Model Performance

library(gmodels)
CrossTable(x = onlinefood_test_labels$Feedback, y=predictions$`k=9`, prop.chisq=FALSE)

## 
##  
##    Cell Contents
## |-------------------------|
## |                       N |
## |           N / Row Total |
## |           N / Col Total |
## |         N / Table Total |
## |-------------------------|
## 
##  
## Total Observations in Table:  78 
## 
##  
##                                 | predictions$`k=9` 
## onlinefood_test_labels$Feedback |  Negative |  Positive | Row Total | 
## --------------------------------|-----------|-----------|-----------|
##                        Negative |        14 |         2 |        16 | 
##                                 |     0.875 |     0.125 |     0.205 | 
##                                 |     0.778 |     0.033 |           | 
##                                 |     0.179 |     0.026 |           | 
## --------------------------------|-----------|-----------|-----------|
##                        Positive |         4 |        58 |        62 | 
##                                 |     0.065 |     0.935 |     0.795 | 
##                                 |     0.222 |     0.967 |           | 
##                                 |     0.051 |     0.744 |           | 
## --------------------------------|-----------|-----------|-----------|
##                    Column Total |        18 |        60 |        78 | 
##                                 |     0.231 |     0.769 |           | 
## --------------------------------|-----------|-----------|-----------|
## 
##

table(predictions$`k=9`, onlinefood_test_labels$Feedback)

##           
##            Negative Positive
##   Negative       14        4
##   Positive        2       58

o_food_knn<-predictions$`k=9`==onlinefood_test_labels$Feedback
o_food_knn

##  [1]  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE
## [13]  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE
## [25]  TRUE  TRUE  TRUE FALSE  TRUE  TRUE  TRUE  TRUE FALSE  TRUE  TRUE  TRUE
## [37]  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE FALSE FALSE  TRUE FALSE  TRUE  TRUE
## [49]  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE
## [61]  TRUE  TRUE  TRUE  TRUE FALSE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE
## [73]  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE

table(o_food_knn)

## o_food_knn
## FALSE  TRUE 
##     6    72

Decision Tree

suppressPackageStartupMessages(library(rattle))
suppressPackageStartupMessages(library(rpart.plot))
library(RColorBrewer)
onlinefood_DT<-online_food

Training & Test Dataset

set.seed(42)
training_indices_DT <- sample(1:nrow(onlinefood_DT), 0.8 * nrow(onlinefood_DT))
traindata_DT <- onlinefood_DT[training_indices_DT, ]
testdata_DT <- onlinefood_DT[-training_indices_DT, ]

Training the model

tree_model_train <- rpart(Feedback ~ ., data = traindata_DT, method = "class")
tree_model_train

## n= 310 
## 
## node), split, n, loss, yval, (yprob)
##       * denotes terminal node
## 
##  1) root 310 54 Positive (0.17419355 0.82580645)  
##    2) Output=No 66 26 Negative (0.60606061 0.39393939)  
##      4) Pin code>=560066.5 16  2 Negative (0.87500000 0.12500000) *
##      5) Pin code< 560066.5 50 24 Negative (0.52000000 0.48000000)  
##       10) Pin code< 560059.5 43 17 Negative (0.60465116 0.39534884)  
##         20) Occupation=Employee,Student 33 10 Negative (0.69696970 0.30303030) *
##         21) Occupation=House wife,Self Employeed 10  3 Positive (0.30000000 0.70000000) *
##       11) Pin code>=560059.5 7  0 Positive (0.00000000 1.00000000) *
##    3) Output=Yes 244 14 Positive (0.05737705 0.94262295) *

fancyRpartPlot(tree_model_train)

Prediction

predictions_DT <- predict(tree_model_train, testdata_DT, type = "class")
predictions_DT

##        1        2        3        4        5        6        7        8 
## Positive Positive Positive Positive Positive Positive Positive Positive 
##        9       10       11       12       13       14       15       16 
## Positive Positive Positive Positive Positive Positive Positive Positive 
##       17       18       19       20       21       22       23       24 
## Positive Positive Positive Positive Negative Negative Positive Positive 
##       25       26       27       28       29       30       31       32 
## Positive Positive Positive Positive Positive Negative Negative Positive 
##       33       34       35       36       37       38       39       40 
## Positive Positive Positive Positive Positive Positive Positive Positive 
##       41       42       43       44       45       46       47       48 
## Negative Positive Negative Positive Negative Negative Positive Positive 
##       49       50       51       52       53       54       55       56 
## Positive Positive Positive Positive Positive Positive Positive Positive 
##       57       58       59       60       61       62       63       64 
## Negative Positive Negative Positive Positive Negative Negative Negative 
##       65       66       67       68       69       70       71       72 
## Positive Positive Positive Negative Positive Positive Positive Negative 
##       73       74       75       76       77       78 
## Positive Positive Positive Negative Negative Positive 
## Levels: Negative Positive

table(predictions_DT)

## predictions_DT
## Negative Positive 
##       17       61

Evaluating Model Performance

library(gmodels)
CrossTable(x=testdata_DT$Feedback, y=predictions_DT, prop.chisq = FALSE)

## 
##  
##    Cell Contents
## |-------------------------|
## |                       N |
## |           N / Row Total |
## |           N / Col Total |
## |         N / Table Total |
## |-------------------------|
## 
##  
## Total Observations in Table:  78 
## 
##  
##                      | predictions_DT 
## testdata_DT$Feedback |  Negative |  Positive | Row Total | 
## ---------------------|-----------|-----------|-----------|
##             Negative |        11 |         6 |        17 | 
##                      |     0.647 |     0.353 |     0.218 | 
##                      |     0.647 |     0.098 |           | 
##                      |     0.141 |     0.077 |           | 
## ---------------------|-----------|-----------|-----------|
##             Positive |         6 |        55 |        61 | 
##                      |     0.098 |     0.902 |     0.782 | 
##                      |     0.353 |     0.902 |           | 
##                      |     0.077 |     0.705 |           | 
## ---------------------|-----------|-----------|-----------|
##         Column Total |        17 |        61 |        78 | 
##                      |     0.218 |     0.782 |           | 
## ---------------------|-----------|-----------|-----------|
## 
##

onlinefood_decisontree<-predictions_DT==testdata_DT$Feedback
onlinefood_decisontree

##  [1]  TRUE  TRUE  TRUE FALSE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE
## [13]  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE FALSE FALSE  TRUE  TRUE
## [25]  TRUE  TRUE  TRUE FALSE  TRUE FALSE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE
## [37]  TRUE  TRUE  TRUE  TRUE FALSE FALSE FALSE  TRUE  TRUE  TRUE FALSE FALSE
## [49]  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE
## [61]  TRUE  TRUE  TRUE  TRUE FALSE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE
## [73]  TRUE  TRUE  TRUE FALSE  TRUE  TRUE

table(onlinefood_decisontree)

## onlinefood_decisontree
## FALSE  TRUE 
##    12    66

prop.table(table(onlinefood_decisontree))

## onlinefood_decisontree
##     FALSE      TRUE 
## 0.1538462 0.8461538

There were 12 incorrectly classified feedbacks and 66 correctly classified feedbacks.

Comparing Machine Learning Models

Support Vector Machine vs KNN vs Decision Tree.

svm_accuracy <- mean(o_food)
knn_accuracy <- mean(o_food_knn)
dt_accuracy <- mean(onlinefood_decisontree)

Accuracy

comparison <- data.frame(
  Model = c("SVM", "KNN", "Decision Tree"),
  Accuracy = c(svm_accuracy, knn_accuracy, dt_accuracy)
)
comparison

##           Model  Accuracy
## 1           SVM 0.8461538
## 2           KNN 0.9230769
## 3 Decision Tree 0.8461538

Accuracy Plot

library(ggplot2)
ggplot(comparison, aes(x = Model, y = Accuracy, fill = Model)) +
  geom_bar(stat = "identity", position = "dodge") +
  labs(title = "Comparison of Model Accuracies", x = "Model", y = "Accuracy") +
  theme_minimal()

Conclusion

Based on the analysis, data preparation and models created, KNN has the best accuracy(92.3%) compared to SVM and Decision Tree.

Online Food Model

Oruru Ndu-Nwogu

2024-05-13

Introduction

Load the Dataset

Analysis

Check for na’s and data types.

Checking Proportions

Standard Deviation

Correlation

Education Qualification Count

Monthly Income

Age Distribution

Feedback based on output

Feedback based on gender

Family Size Distribution

Data Preparation

Factors

Dependent Variable

Support Vector Machine

Normalization

Create Training and Test Dataset

Create Classifier

Evaluate Onlinefood Prediction

Understanding Classification Results

Improving model performance

Prediction

Classification Results

Confusion Matrix

KNN

Data Preparation

Normalization

Training and Test Dataset

Training & Test (No label)

Training & Test (Labelled)

Choosing best k-value

Visualizing k-values

Evaluate Model Performance

Decision Tree

Training & Test Dataset

Training the model

Prediction

Evaluating Model Performance

Comparing Machine Learning Models

Accuracy

Accuracy Plot

Conclusion