Project Description

• Author: Bhishan Poudel
• Project: Prudential Life Insurance policy prediction (Company)
• Kaggle project page: Prudential Life Insurance Assessment
• Task: Multiclass classification (Response 1 to 8)
• Metric of evaluation: Quadratic Weighted Kappa

Data

Here Response 8 has the highest counts, I assume it the quote that is granted.

Records: 60k
Features: 127
Target: Response (has 8 categories, 1-8)

Features:

1 Misc             : Age ht wt bmi              4
2 Product Info     : Product_Info_1 to 7        7
3 Employment Info  : Employment_Info_1 to 6     6
4 Insured Info     : InsuredInfo_1 to 7         7
5 Insurance History: Insurance_History_1 to 9   9
6 Family History   : Family_Hist_1 to 5         5
7 Medical History  : Medical_History_1 to 41    41
8 Medical Keywords : Medical_Keyword_1 to 48    48
Target: Response                                1
ID    : ID                                      1
---------------------------------------------------
Total Features: 127
Dependent Variable: 1 (Response)

Metric of Evaluation

Quadratic weighted kappa is rank based metric used for multiclass classification. It has the minimum value 0 (random guess) and maximum value 1 (total agreement).

from sklearn import metrics
score = metrics.cohen_kappa_score(ytest,ypreds,weights='quadratic')

Exploratory Data Analysis

Modelling

I have tried various machine learning models for this project. The results are presented below.

Linear Regression

Weighted quadratic kappa = 0.569630318923444

Xgboost Classifier softprob

objective = "multi:softprob"

Cleaning   eval_metric   kappa
simple     default       0.5377897181694622
detailed   default       0.5438646999324421
simple     custom        0.5530053047703208
detailed   custom        0.5407784634778012

Xgboost poission regressor

objective = "count:poisson"

   Model                    TrainKappa  TestKappa
0   xgb reg                 0.669651      0.603765
1   xgb reg + offset        0.720368      0.649496
2   xgb poisson             0.682188      0.609387
3   xgb poisson + offset    0.735050      0.655627

Xgboost poisson regressor ensemble voting

    Model                   TrainKappa     TestKappa
0   xgb reg                 0.669651         0.603765
1   xgb reg + offset        0.720368         0.649496
2   xgb poisson             0.682188         0.609387
3   xgb poisson + offset    0.735050         0.655627
4   ensemble                0.623919         0.593312
5   ensemble + offset       0.683268         0.644076

Model Explanation