sarah-2510 / vehicle-insurance-claim-prediction

A model was built to predict the total insurance claim amount payable by the insurance company using machine learning techniques such as regression in python.

Vehicle-Insurance-Claim-Prediction

File Description

• Code: The code used to work on the dataset
• Test_SJC: The testing dataset
• train_SJC: The training dataset

Data Description

• ClaimNumber : Unique policy identifier
• DateTimeOfAccident : Date and time of accident
• DateReported : Date that accident was reported
• Age : Age of worker
• Gender : Gender of worker
• MaritalStatus : Martial status of worker. (M)arried, (S)ingle, (U)unknown.
• DependentChildren : The number of dependent children
• DependentsOther : The number of dependants excluding children
• WeeklyWages : Total weekly wage
• PartTimeFullTime : Binary (P) or (F)
• HoursWorkedPerWeek : Total hours worked per week
• DaysWorkedPerWeek : Number of days worked per week
• ClaimDescription : Free text description of the claim
• InitialIncurredClaimCost : Initial estimate by the insurer of the claim cost
• UltimateIncurredClaimCost : Total claims payments by the insurance company. This is the field you are asked to predict in the test set.

Problem Statement

To predict the Total Claims Payments by the insurance company.

The approach to work on the dataset in as follows:

1. Importing the libraries that was needed.

2. Pre-Processing

In the pre-processing I wrote codes to obtain basic information on the dataset such as data description,data shape, whether or not if there are any duplicate values or missing data in the dataset. As some of the columns had some missing data I treated them using mean and mode imputation. Following that I worked on binning some of the data columns into small intervals and replaced it by a general value calculated for that bin . This is done in order to understand those columns further more and analyse them.

3. Exploratory Data Analysis

After data pre-processing I focused on exploring the data.This was done in three parts

i) Univariate Analysis

ii) Bi-variate Analysis

iii) Multi-variate Analysis

I examined the link between different variables using histogram(for numerical variables) and countplots(for categorical variables). Using Correlation plots, I tried to understand the relationship between all the numerical variables with each other as well as the target variable.

4. Outlier Analysis

Using boxplots, I looked for outliers. Outliers were handled using a function.

5. Machine Learning Models

I experimented with three different models: Linear regression, Decision tree regression, and Random forest regression.

6. Result

The RSME score for all the models were as follows

i) Linear regression : 1907

ii) Decision Tree regressor : 2407

iii) Random forest Regressor : 1799

Out of these 3 I selected random forest regressor as it was giving the least root mean squared error (RMSE).