Basic Statistics

The training data has 8,161 rows and 25 columns or features. Of the 25 columns, 14 are discrete, 11 are continuous, and none are all their missing. We do have 970 missing values out of 204,025 data points.

Distribution of Target Variable

Let’s look at the target_flag variable in our training data to make sure there no one sided distribution.

## [1] 6008

Histogram of Variables

Boxplot of Variables

## NULL

Adjusting Variables

Looking at the plots we see we have to make a few changes to some variables. We’ll make HOMEKIDS boolean instead of a factor. For the rows where CAR_AGE are labled numbers, we make 0. For blank JOBS we label those as “Unknown”. Finally, change Education to 1 if PhD and Masters.

Missing Data

We have missing data in income, yoj, home_val, and car_age.

##  TARGET_FLAG   TARGET_AMT        KIDSDRIV           AGE       
##  0:6008      Min.   :     0   Min.   :0.0000   Min.   :16.00  
##  1:2153      1st Qu.:     0   1st Qu.:0.0000   1st Qu.:39.00  
##              Median :     0   Median :0.0000   Median :45.00  
##              Mean   :  1504   Mean   :0.1711   Mean   :44.79  
##              3rd Qu.:  1036   3rd Qu.:0.0000   3rd Qu.:51.00  
##              Max.   :107586   Max.   :4.0000   Max.   :81.00  
##                                                NA's   :6      
##     HOMEKIDS           YOJ           INCOME       PARENT1   
##  Min.   :0.0000   Min.   : 0.0   Min.   :     0   No :7084  
##  1st Qu.:0.0000   1st Qu.: 9.0   1st Qu.: 28097   Yes:1077  
##  Median :0.0000   Median :11.0   Median : 54028             
##  Mean   :0.3519   Mean   :10.5   Mean   : 61898             
##  3rd Qu.:1.0000   3rd Qu.:13.0   3rd Qu.: 85986             
##  Max.   :1.0000   Max.   :23.0   Max.   :367030             
##                   NA's   :454    NA's   :445                
##     HOME_VAL      MSTATUS      SEX         EDUCATION     
##  Min.   :     0   Yes :4894   M  :3786   Min.   :0.0000  
##  1st Qu.:     0   z_No:3267   z_F:4375   1st Qu.:0.0000  
##  Median :161160                          Median :1.0000  
##  Mean   :154867                          Mean   :0.7076  
##  3rd Qu.:238724                          3rd Qu.:1.0000  
##  Max.   :885282                          Max.   :1.0000  
##  NA's   :464                                             
##             JOB          TRAVTIME            CAR_USE        BLUEBOOK    
##  z_Blue Collar:1825   Min.   :  5.00   Commercial:3029   Min.   : 1500  
##  Clerical     :1271   1st Qu.: 22.00   Private   :5132   1st Qu.: 9280  
##  Professional :1117   Median : 33.00                     Median :14440  
##  Manager      : 988   Mean   : 33.49                     Mean   :15710  
##  Lawyer       : 835   3rd Qu.: 44.00                     3rd Qu.:20850  
##  Student      : 712   Max.   :142.00                     Max.   :69740  
##  (Other)      :1413                                                     
##       TIF                CAR_TYPE    RED_CAR       OLDCLAIM    
##  Min.   : 1.000   Minivan    :2145   no :5783   Min.   :    0  
##  1st Qu.: 1.000   Panel Truck: 676   yes:2378   1st Qu.:    0  
##  Median : 4.000   Pickup     :1389              Median :    0  
##  Mean   : 5.351   Sports Car : 907              Mean   : 4037  
##  3rd Qu.: 7.000   Van        : 750              3rd Qu.: 4636  
##  Max.   :25.000   z_SUV      :2294              Max.   :57037  
##                                                                
##     CLM_FREQ      REVOKED       MVR_PTS          CAR_AGE      
##  Min.   :0.0000   No :7161   Min.   : 0.000   Min.   : 0.000  
##  1st Qu.:0.0000   Yes:1000   1st Qu.: 0.000   1st Qu.: 1.000  
##  Median :0.0000              Median : 1.000   Median : 8.000  
##  Mean   :0.7986              Mean   : 1.696   Mean   : 8.329  
##  3rd Qu.:2.0000              3rd Qu.: 3.000   3rd Qu.:12.000  
##  Max.   :5.0000              Max.   :13.000   Max.   :28.000  
##                                               NA's   :510     
##                  URBANICITY  
##  Highly Urban/ Urban  :6492  
##  z_Highly Rural/ Rural:1669  
##                              
##                              
##                              
##                              
## 

Imputate

Let’s runmice imputation on both the train and test set.

Model 1

For the first model we will include all the variables. Looking at the output of the model we see that some points are highly colinear and a some variables that may not be necessary.

Model 1 uses the formula:

target ~ .

Model 2

For the second model we ignore what’s colinear but remove unneccessary variables shown in model 1.

Model 2 uses the formula:

TARGET_FLAG ~ KIDSDRIV + INCOME + PARENT1 + HOME_VAL + MSTATUS + JOB + TRAVTIME + CAR_USE + BLUEBOOK + TIF + CAR_TYPE + OLDCLAIM + CLM_FREQ + REVOKED + MVR_PTS + URBANICITY

Model 3

Model 3 removes the variables with the 3 highest VIF values from Model 1.

Model 3 uses the formula:

TARGET_FLAG ~ KIDSDRIV + AGE + HOMEKIDS + YOJ + INCOME + PARENT1 + HOME_VAL + MSTATUS + JOB + TRAVTIME + CAR_USE + BLUEBOOK + TIF + CAR_TYPE + RED_CAR + OLDCLAIM + CLM_FREQ + REVOKED + MVR_PTS + CAR_AGE + URBANICITY

Model 4

Model 4 takes in Model 3 and removes thoe values that are poor predictors.

Model 4 uses the formula:

TARGET_FLAG ~ KIDSDRIV + PARENT1 + HOME_VAL + MSTATUS + JOB + TRAVTIME + CAR_USE + BLUEBOOK + TIF + CAR_TYPE + OLDCLAIM + CLM_FREQ + REVOKED + MVR_PTS + CAR_AGE + URBANICITY

1: Classification

Fourfold Plots

Summary Statistics

## 
## Call:
## NULL
## 
## Deviance Residuals: 
##     Min       1Q   Median       3Q      Max  
## -2.5844  -0.7208  -0.4001   0.6454   3.1284  
## 
## Coefficients:
##                                   Estimate Std. Error z value Pr(>|z|)    
## (Intercept)                       -1.42296    0.03476 -40.938  < 2e-16 ***
## KIDSDRIV                           0.21112    0.02805   7.527 5.18e-14 ***
## PARENT1Yes                         0.15818    0.03183   4.970 6.69e-07 ***
## HOME_VAL                          -0.24811    0.04056  -6.117 9.56e-10 ***
## MSTATUSz_No                        0.18048    0.03858   4.679 2.89e-06 ***
## JOBDoctor                         -0.17125    0.04001  -4.280 1.87e-05 ***
## `JOBHome Maker`                   -0.03334    0.03442  -0.969 0.332706    
## JOBLawyer                         -0.14305    0.04208  -3.400 0.000674 ***
## JOBManager                        -0.38541    0.04229  -9.114  < 2e-16 ***
## JOBProfessional                   -0.16704    0.03915  -4.266 1.99e-05 ***
## JOBStudent                        -0.03436    0.03509  -0.979 0.327509    
## JOBUnknown                        -0.14275    0.03832  -3.726 0.000195 ***
## `JOBz_Blue Collar`                -0.07207    0.04373  -1.648 0.099379 .  
## TRAVTIME                           0.22459    0.02982   7.530 5.06e-14 ***
## CAR_USEPrivate                    -0.34960    0.04189  -8.346  < 2e-16 ***
## BLUEBOOK                          -0.22306    0.03903  -5.715 1.10e-08 ***
## TIF                               -0.22632    0.03035  -7.458 8.81e-14 ***
## `CAR_TYPEPanel Truck`              0.17693    0.04107   4.308 1.65e-05 ***
## CAR_TYPEPickup                     0.21474    0.03742   5.738 9.56e-09 ***
## `CAR_TYPESports Car`               0.30308    0.03369   8.997  < 2e-16 ***
## CAR_TYPEVan                        0.18735    0.03505   5.346 9.01e-08 ***
## CAR_TYPEz_SUV                      0.31923    0.03855   8.281  < 2e-16 ***
## OLDCLAIM                          -0.12166    0.03425  -3.552 0.000382 ***
## CLM_FREQ                           0.22721    0.03294   6.898 5.29e-12 ***
## REVOKEDYes                         0.29136    0.02982   9.772  < 2e-16 ***
## MVR_PTS                            0.24830    0.02909   8.535  < 2e-16 ***
## CAR_AGE                           -0.11795    0.03659  -3.223 0.001267 ** 
## `URBANICITYz_Highly Rural/ Rural` -0.95632    0.04534 -21.091  < 2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for binomial family taken to be 1)
## 
##     Null deviance: 9418.0  on 8160  degrees of freedom
## Residual deviance: 7329.4  on 8133  degrees of freedom
## AIC: 7385.4
## 
## Number of Fisher Scoring iterations: 5

Make Predictions

Finally we make our final predictions. We’ll create a dataframe with the predictions and the predicted probabilities for our classification.

For our predictive model we’re off on the classification. We’ll make a prediction on the target amount for all observations in the test set, regardless of whether we think they’ll make a claim.