R Notebook

cars <- read.csv("mode_of_transport.csv")
View(cars)

Structure of the data

str(cars)

'data.frame':   444 obs. of  9 variables:
 $ Age      : int  28 23 29 28 27 26 28 26 22 27 ...
 $ Gender   : chr  "Male" "Female" "Male" "Female" ...
 $ Engineer : int  0 1 1 1 1 1 1 1 1 1 ...
 $ MBA      : int  0 0 0 1 0 0 0 0 0 0 ...
 $ Work.Exp : int  4 4 7 5 4 4 5 3 1 4 ...
 $ Salary   : num  14.3 8.3 13.4 13.4 13.4 12.3 14.4 10.5 7.5 13.5 ...
 $ Distance : num  3.2 3.3 4.1 4.5 4.6 4.8 5.1 5.1 5.1 5.2 ...
 $ license  : int  0 0 0 0 0 1 0 0 0 0 ...
 $ Transport: chr  "Public Transport" "Public Transport" "Public Transport" "Public Transport" ...

Variables like Engineer, MBA, license contains category values. So, we need to convert them into factors.

cars$Engineer = as.factor(cars$Engineer)
cars$MBA = as.factor(cars$MBA)
cars$license = as.factor(cars$license)
cars$Gender = as.factor(cars$Gender)
cars$Transport = as.factor(cars$Transport)

Summary of dataset

summary(cars)

      Age           Gender    Engineer MBA        Work.Exp        Salary     
 Min.   :18.00   Female:128   0:109    0:332   Min.   : 0.0   Min.   : 6.50  
 1st Qu.:25.00   Male  :316   1:335    1:112   1st Qu.: 3.0   1st Qu.: 9.80  
 Median :27.00                                 Median : 5.0   Median :13.60  
 Mean   :27.75                                 Mean   : 6.3   Mean   :16.24  
 3rd Qu.:30.00                                 3rd Qu.: 8.0   3rd Qu.:15.72  
 Max.   :43.00                                 Max.   :24.0   Max.   :57.00  
    Distance     license             Transport  
 Min.   : 3.20   0:340   Private Transport:144  
 1st Qu.: 8.80   1:104   Public Transport :300  
 Median :11.00                                  
 Mean   :11.32                                  
 3rd Qu.:13.43                                  
 Max.   :23.40                                  

Here, the minimum age is 18, maximum age is 43 and average age is 27. We conclude that the majority is of males of approximately 70-77%. Here, there are more Engineers than MBA’s. And, the most common mode of transport is public transportation.

Visual Analysis of data

boxplot(cars$Age ~ cars$Engineer, main="Engineer's Age")

boxplot(cars$Age ~ cars$MBA, main="MBA's Age")

There’s no much difference in the age of Engineers and MBA’s.

Let’s find differences in the average salary for two professions.

boxplot(cars$Salary ~ cars$Engineer, main="Engineer's Salary")

boxplot(cars$Salary ~ cars$MBA, main="MBA's Salary")

There’s isn’t any major difference in the salary.

Let’s see the Work Exp column

hist(cars$Work.Exp, main="Distribution of Work Experience")

This column is right skewed, this is expected as there would be more juniors then seniors in any firm.

boxplot(cars$Work.Exp ~ cars$Gender)

The box plot shows that there is an equal distribution of male and female with no such difference in work experience.

table(cars$license, cars$Transport)

   
    Private Transport Public Transport
  0                73              267
  1                71               33

The table clearly shows that those who don’t have license uses public transport more.

boxplot(cars$Salary ~ cars$Transport, main="Mode of transport according to the salary")

Plot clearly shows that as the salary increases the number of individuals with private transport also increases.

Let’s see if the age and work experience can be usage of private transport increases

cor(cars$Age, cars$Work.Exp)

[1] 0.9322364

boxplot(cars$Age ~ cars$Transport, main="Age and Transport")

The plot clearly explains that the higher age prefers private transport.

Let’s see with the increase in distance would the employees prefer private or public transport.

boxplot(cars$Distance ~ cars$Transport, main="Preference of transport according to distance")

In this plot the pattern shows that for longer distance private transport is the preference of employees.

Let’s see whether female employees prefer private or public transport

table(cars$Gender, cars$Transport)

        
         Private Transport Public Transport
  Female                51               77
  Male                  93              223

From the above data it is visible that more females prefers Public Transport.

Only 40% female uses private transport and only 30% male uses private transport.

Data Preparations

Let’s find if there is any missing values in the data set.

anyNA(cars)

[1] FALSE

set.seed(1234)
pd = sample(2, nrow(cars), replace=TRUE, prob = c(0.7, 0.3))
carstrain = cars[pd==1,]
carstest = cars[pd==2,]
carstest$Salary = log(carstest$Salary)
carstest$Engineer = as.factor(carstest$Engineer)
carstest$MBA = as.factor(carstest$MBA)
carstest$license = as.factor(carstest$license)

There isn’t any missing values in this data set.

Arrange Test Set

library(caret)
random <- createDataPartition(cars$Transport, p=0.70, list=FALSE)

train_set = cars[random,]
test_set = cars[-random,]

Model Building

Naive Bayes

library(e1071)
nb_model <- naiveBayes(train_set$Transport ~ ., data = train_set)

nb_model

Naive Bayes Classifier for Discrete Predictors

Call:
naiveBayes.default(x = X, y = Y, laplace = laplace)

A-priori probabilities:
Y
Private Transport  Public Transport 
        0.3247588         0.6752412 

Conditional probabilities:
                   Age
Y                       [,1]     [,2]
  Private Transport 29.31683 5.847958
  Public Transport  26.72381 3.081028

                   Gender
Y                      Female      Male
  Private Transport 0.3861386 0.6138614
  Public Transport  0.2428571 0.7571429

                   Engineer
Y                           0         1
  Private Transport 0.2079208 0.7920792
  Public Transport  0.2190476 0.7809524

                   MBA
Y                           0         1
  Private Transport 0.7920792 0.2079208
  Public Transport  0.6952381 0.3047619

                   Work.Exp
Y                       [,1]     [,2]
  Private Transport 8.326733 6.855813
  Public Transport  4.952381 3.220392

                   Salary
Y                       [,1]      [,2]
  Private Transport 21.67624 14.923787
  Public Transport  13.19238  4.956069

                   Distance
Y                       [,1]     [,2]
  Private Transport 13.45545 3.483345
  Public Transport  10.39000 2.915762

                   license
Y                           0         1
  Private Transport 0.5346535 0.4653465
  Public Transport  0.8857143 0.1142857

nb_predictions <- predict(nb_model, test_set)
table(nb_predictions, test_set$Transport)

                   
nb_predictions      Private Transport
  Private Transport                21
  Public Transport                 22
                   
nb_predictions      Public Transport
  Private Transport                3
  Public Transport                87

Logistic Regression

logistic_model <- glm(Transport~., data = train_set, family = binomial(link = "logit"))
test_set$log.pred <- predict(logistic_model, test_set[1:8], type="response")

table(test_set$Transport, test_set$log.pred>0.5)

                   
                    FALSE TRUE
  Private Transport    24   19
  Public Transport      8   82

KNN

trControl <- trainControl(method = "cv", number = 10)
fit.knn <- train(Transport ~ ., method = "knn",
                 tuneGrid = expand.grid(k=2:20),
                 trControl=trControl,
                 metric = "Accuracy",
                 preProcess = c("center", "scale"),
                 data = train_set)
fit.knn

k-Nearest Neighbors 

311 samples
  8 predictor
  2 classes: 'Private Transport', 'Public Transport' 

Pre-processing: centered (8), scaled (8) 
Resampling: Cross-Validated (10 fold) 
Summary of sample sizes: 280, 280, 280, 280, 280, 280, ... 
Resampling results across tuning parameters:

  k   Accuracy   Kappa    
   2  0.7617944  0.4418982
   3  0.8039315  0.5306667
   4  0.7913306  0.5027636
   5  0.7912298  0.4954522
   6  0.7750000  0.4515462
   7  0.7910282  0.4862478
   8  0.7748992  0.4463433
   9  0.8007056  0.4927173
  10  0.7784274  0.4417140
  11  0.7847782  0.4481347
  12  0.7814516  0.4294389
  13  0.7911290  0.4533349
  14  0.8104839  0.5020040
  15  0.8007056  0.4790518
  16  0.7878024  0.4438665
  17  0.7813508  0.4223460
  18  0.7782258  0.4135288
  19  0.7781250  0.4167973
  20  0.7717742  0.3926983

Accuracy was used to select the optimal
 model using the largest value.
The final value used for the model was k = 14.

knn_pred_train <- predict(fit.knn, train_set)

table(knn_pred_train, train_set$Transport)

                   
knn_pred_train      Private Transport
  Private Transport                54
  Public Transport                 47
                   
knn_pred_train      Public Transport
  Private Transport                7
  Public Transport               203

knn_pred_test <- predict(fit.knn, test_set)
table(knn_pred_test, test_set$Transport)

                   
knn_pred_test       Private Transport
  Private Transport                23
  Public Transport                 20
                   
knn_pred_test       Public Transport
  Private Transport                2
  Public Transport                88

predict(fit.knn, carstest)

  [1] Public Transport  Public Transport 
  [3] Public Transport  Public Transport 
  [5] Public Transport  Public Transport 
  [7] Public Transport  Public Transport 
  [9] Public Transport  Public Transport 
 [11] Public Transport  Public Transport 
 [13] Public Transport  Public Transport 
 [15] Public Transport  Public Transport 
 [17] Public Transport  Public Transport 
 [19] Public Transport  Public Transport 
 [21] Public Transport  Public Transport 
 [23] Public Transport  Public Transport 
 [25] Public Transport  Public Transport 
 [27] Public Transport  Public Transport 
 [29] Public Transport  Public Transport 
 [31] Public Transport  Public Transport 
 [33] Public Transport  Public Transport 
 [35] Public Transport  Public Transport 
 [37] Public Transport  Public Transport 
 [39] Public Transport  Public Transport 
 [41] Public Transport  Public Transport 
 [43] Public Transport  Public Transport 
 [45] Private Transport Public Transport 
 [47] Public Transport  Public Transport 
 [49] Private Transport Public Transport 
 [51] Public Transport  Public Transport 
 [53] Private Transport Public Transport 
 [55] Public Transport  Public Transport 
 [57] Public Transport  Public Transport 
 [59] Public Transport  Public Transport 
 [61] Public Transport  Public Transport 
 [63] Public Transport  Public Transport 
 [65] Private Transport Public Transport 
 [67] Public Transport  Public Transport 
 [69] Public Transport  Private Transport
 [71] Public Transport  Private Transport
 [73] Public Transport  Public Transport 
 [75] Private Transport Public Transport 
 [77] Private Transport Public Transport 
 [79] Public Transport  Public Transport 
 [81] Public Transport  Public Transport 
 [83] Public Transport  Public Transport 
 [85] Public Transport  Public Transport 
 [87] Public Transport  Public Transport 
 [89] Public Transport  Public Transport 
 [91] Public Transport  Public Transport 
 [93] Public Transport  Private Transport
 [95] Public Transport  Public Transport 
 [97] Public Transport  Public Transport 
 [99] Public Transport  Private Transport
[101] Private Transport Public Transport 
[103] Private Transport Public Transport 
[105] Private Transport Private Transport
[107] Public Transport  Public Transport 
[109] Public Transport  Private Transport
[111] Private Transport Private Transport
[113] Private Transport Public Transport 
[115] Private Transport Private Transport
[117] Private Transport Public Transport 
[119] Public Transport  Private Transport
[121] Private Transport Public Transport 
[123] Public Transport  Public Transport 
[125] Private Transport Private Transport
[127] Private Transport Private Transport
[129] Private Transport Private Transport
[131] Private Transport Private Transport
[133] Private Transport
Levels: Private Transport Public Transport

Conclusion

• Important variables are Age, Work Exp, Distance and License
• Age and Work Exp are highly correlated
• Employees with work exp more than 10 years and above are likely to use private transport
• Employees who must commute for distance greater than 12 are more likely to use private transport
• People with higher salaries greater than 20k are most likely to use private transportation
• With license, we do see that 81% who commute through private transportation have license and 89% commute through public transport