本项目旨在分析基于区域、产品类别、索赔金额等因素的保修索赔数据，以预测索赔的真实性数据集来源于Kaggle，包含358条记录和资源-CSDN文库

共4个文件

md：1个

pdf：1个

ipynb：1个

数据集

122 浏览量 2024-07-21 16:37:01 上传评论收藏 1.48MB ZIP 举报

资源推荐

资源详情

资源评论

收起资源包目录

Warranty Claims Fraud Prediction.zip （4个子文件）

df_Clean.csv 38KB

Warranty Claims Fraud Prediction.pdf 1.37MB

description.md 4KB

Warranty Claims Fraud Prediction.ipynb 1.06MB

9/7/23, 9:33 AM

Warranty Claims Fraud Prediction

file:///E:/Data Science Course/Projects/Warranty Claims Fraud Prediction.html

1/20

Warranty Claims Fraud Prediction

The aim of this project is to analyze the warranty claims based on their region, product,

claim value and other features to predict their authenticity. The dataset is taken from

Kaggle. The dataset contains 358 rows and 21 columns.

Data Dictionary

Column Name Description

Unnamed: 0 Index

Region Region of the claim

State State of the claim

Area Area of the claim

City City of the claim

Consumer_profile Consumer profile Business/Personal

Product_category Product category Household/Entertainment

Product_type Product type AC/TV

AC_1001_Issue 1 0- No issue / No componenent, 1- repair, 2-replacement

AC_1002_Issue 1 0- No issue / No componenent, 1- repair, 2-replacement

AC_1003_Issue 1 0- No issue / No componenent, 1- repair, 2-replacement

TV_2001_Issue 1 0- No issue / No componenent, 1- repair, 2-replacement

TV_2002_Issue 1 0- No issue / No componenent, 1- repair, 2-replacement

TV_2003_Issue 1 0- No issue / No componenent, 1- repair, 2-replacement

Claim_Value Claim value in INR

Service_Center Service center code

Product_Age Product age in days

Purchased_from Purchased from - Dealer, Manufacturer, Internet

Call_details Call duration

Purpose Purpose of the call

Fraud Fraudulent (1) or Genuine (0)

In [ ]:

# Importing the libraries

import numpy as np

import matplotlib.pyplot as plt

import pandas as pd

import seaborn as sns

9/7/23, 9:33 AM

Warranty Claims Fraud Prediction

file:///E:/Data Science Course/Projects/Warranty Claims Fraud Prediction.html

2/20

Unnamed:

Region State Area City Consumer_profile Product_categor

0 0 South Karnataka Urban Bangalore Business Entertainmen

1 1 South Karnataka Rural Bangalore Business Househo

2 2 North Haryana Urban Chandigarh Personal Househo

3 3 South

Tamil

Nadu

Urban Chennai Business Entertainmen

4 4

North

East

Jharkhand Rural Ranchi Personal Entertainmen

5 rows × 21 columns

Data Preprocessing Part 1

(358, 21)

Region 0

State 0

Area 0

City 0

Consumer_profile 0

Product_category 0

Product_type 0

AC_1001_Issue 0

AC_1002_Issue 0

AC_1003_Issue 0

TV_2001_Issue 0

TV_2002_Issue 0

TV_2003_Issue 0

Claim_Value 0

Service_Centre 0

Product_Age 0

Purchased_from 0

Call_details 0

Purpose 0

Fraud 0

dtype: int64

In [ ]:

#Loading the dataset

df = pd.read_csv('df_Clean.csv')

df.head()

Out[ ]:

In [ ]:

# checking the shape of the dataset

df.shape

Out[ ]:

In [ ]:

# Drop index column

df.drop(['Unnamed: 0'], axis=1, inplace=True)

In [ ]:

# Checking for null/missing values

df.isnull().sum()

Out[ ]:

9/7/23, 9:33 AM

Warranty Claims Fraud Prediction

file:///E:/Data Science Course/Projects/Warranty Claims Fraud Prediction.html

3/20

Region object

State object

Area object

City object

Consumer_profile object

Product_category object

Product_type object

AC_1001_Issue int64

AC_1002_Issue int64

AC_1003_Issue int64

TV_2001_Issue int64

TV_2002_Issue int64

TV_2003_Issue int64

Claim_Value float64

Service_Centre int64

Product_Age int64

Purchased_from object

Call_details float64

Purpose object

Fraud int64

dtype: object

Region 8

State 20

Area 2

City 27

Consumer_profile 2

Product_category 2

Product_type 2

AC_1001_Issue 3

AC_1002_Issue 3

AC_1003_Issue 3

TV_2001_Issue 3

TV_2002_Issue 3

TV_2003_Issue 3

Claim_Value 107

Service_Centre 7

Product_Age 188

Purchased_from 3

Call_details 37

Purpose 3

Fraud 2

dtype: int64

In [ ]:

# Checking for duplicate values

df.duplicated().sum()

Out[ ]:

In [ ]:

# Checking the data types

df.dtypes

Out[ ]:

In [ ]:

# Unique values in each column

df.nunique()

Out[ ]:

In [ ]:

# renaming the values in product issue column

df['AC_1001_Issue'] = df['AC_1001_Issue'].map({ 0 : 'No Issue', 1 : 'repair', 2

df['AC_1002_Issue'] = df['AC_1002_Issue'].map({ 0 : 'No Issue', 1 : 'repair', 2

df['AC_1003_Issue'] = df['AC_1003_Issue'].map({ 0 : 'No Issue', 1 : 'repair', 2

9/7/23, 9:33 AM

Warranty Claims Fraud Prediction

file:///E:/Data Science Course/Projects/Warranty Claims Fraud Prediction.html

4/20

Descriptive Statistics

Claim_Value Service_Centre Product_Age Call_details Fraud

count 358.000000 358.000000 358.000000 358.000000 358.000000

mean 11994.534916 12.812849 201.843575 11.931844 0.097765

std 12063.213579 1.766844 259.731564 11.559474 0.297413

min 0.000000 10.000000 3.000000 0.500000 0.000000

25% 4006.000000 12.000000 14.000000 1.600000 0.000000

50% 7194.000000 13.000000 60.000000 6.500000 0.000000

75% 15000.000000 15.000000 303.750000 23.000000 0.000000

max 50000.000000 16.000000 991.000000 30.000000 1.000000

Region State Area City Consumer_profile Product_category Product_

0 South Karnataka Urban Bangalore Business Entertainment

1 South Karnataka Rural Bangalore Business Household

2 North Haryana Urban Chandigarh Personal Household

3 South

Tamil

Nadu

Urban Chennai Business Entertainment

North

East

Jharkhand Rural Ranchi Personal Entertainment

Exploratory Data Analysis

Location based Distribution of Fraudulent Claims

df['TV_2001_Issue'] = df['TV_2001_Issue'].map({ 0 : 'No Issue', 1 : 'repair', 2

df['TV_2002_Issue'] = df['TV_2002_Issue'].map({ 0 : 'No Issue', 1 : 'repair', 2

df['TV_2003_Issue'] = df['TV_2003_Issue'].map({ 0 : 'No Issue', 1 : 'repair', 2

In [ ]:

df.describe()

Out[ ]:

In [ ]:

df.head()

Out[ ]:

In [ ]:

fig, ax = plt.subplots(2,2,figsize=(15,10))

fig.subplots_adjust(hspace=0.7)

sns.histplot(x = 'Region', data = df, ax =ax[0,0], hue = 'Fraud', element='bars'

ax[0,0].xaxis.set_tick_params(rotation=90)

sns.histplot(x = 'State', data = df, ax =ax[0,1], hue = 'Fraud', element='bars',

ax[0,1].xaxis.set_tick_params(rotation=90)

sns.histplot(x = 'City', data = df, ax =ax[1,0], hue = 'Fraud', element='bars',

9/7/23, 9:33 AM

Warranty Claims Fraud Prediction

file:///E:/Data Science Course/Projects/Warranty Claims Fraud Prediction.html

5/20

[Text(0.5, 1.0, 'Areawise Distribution of Fraudulent Claims')]

The above plots visualizes the distribution of fraudulent claims based on location. The

first graphs shpws the regional distribution of the fraudent claims, where South, North

East and south East are among the regionas with highest warranty claims. However, the

regions - West, East and South are among regions with highest fraudulent claims.

Interestingly the North West region has zero fraudent claims.

The second graph shows the distribution of fraudulent claims based on the States, where

the states - Andhra Pradesh, Maharashtra, Tamil Nadu, Karnataka and Gujarat are among

the states with highest number of warranty claims and states - Haryana has lowest

warranty claims. The states - Andhra Pradesh, Maharashtra, Tamil Nadu, Karnataka and

Gujarat are among the states with highest number of fraudulent claims whereas, states

like Bihar, Delhi, West Bengal and Assam are among the states with lowest number of

fraudulent claims.

The third graph shows the distribution of fraudulent claims based on cities. The cities -

Chennai, Hyderabad, Bangalore, Mumbai and Kochi are among the cities with highest

claims whereas cities like Chandigarh, Srinagar, Agartala and Shimla have lowest number

of claims. Moreover the cities - Chennai, Hyderabad, Bangalore, Mumbai and Kochi are

among the cities with highest fraudulent claims whereas cities like Chandigarh, Panaji,

Meerut, Jaipur, and many other have zero fraudulent claims.

ax[1,0].xaxis.set_tick_params(rotation=90)

sns.histplot(x = 'Area', data = df, ax =ax[1,1], hue = 'Fraud', element='bars',

Out[ ]:

评论收藏

内容反馈

小夕Coding

粉丝: 6202
资源: 526

本项目旨在分析基于区域、产品类别、索赔金额等因素的保修索赔数据，以预测索赔的真实性 数据集来源于Kaggle，包含358条记录和

对 Kaggle 上可用的“美国事故”数据集进行探索性数据分析

kaggle-insurance-claim-classification:该项目旨在建立一个预测模型，该模型可以预测特定索赔将被立即批准或不被批准的概率

Python实现保险公司的理赔支出预测源码+数据集， 健康保险理赔预测源码，对特征数据进行可视化分析，用线性回归模型进行机器学习

allstate-claims-severity.zip

kaggle_bnp:法国巴黎银行Cardif索赔管理

loss-in-insurance-policy

Porto_Seguro’s_Safe_Driver_Prediction.zip

DGVFM_KI:DGVFM-AI-Broschure的存储库

小猿口算自动化答题，让他明白什么叫做数学天才

python3.12对应的dlib-19.24.99-cp312-cp312-win-amd64

《点燃我温暖你》中李峋的同款爱心代码

Python入门基础教程全套.ppt

Google Chrome浏览器ChromeDriver驱动下载(Chrome版本：130.0.6723.59)win64

Google Chrome浏览器ChromeDriver驱动下载(Chrome版本：129.0.6668.101)win64

PUBG罗技宏代码免费

小猿口算各种脚本（OCR+模拟正确答案，劫持数据包+模拟正确答案，劫持数据包+篡改正确答案）（源码）

基于Python的小猿口算OCR识别比大小脚本（源码）

Python学习笔记(干货) 中文PDF完整版.pdf

PyCharm 激活方法

Tesseract最新中文语言包chi-sim.traineddata

yolov8源码+yolov8n、s、m.pt文件整合8.2.0版本

122版本Chrome最新驱动-122.0.6261.58

微信小程序+反编译工具

Microsoft C++ Build Tools

libomp140.x86-64.dll

小猿口算连点器脚本,1秒10道题, 解压有软件和视频教程

华为商城抢购huawei mate xt非凡大师脚本

国家中小学智慧教育平台（课件、课本、视频 ）下载器

小猿口算比大小速算脚本.rar

最新资源

本项目旨在分析基于区域、产品类别、索赔金额等因素的保修索赔数据，以预测索赔的真实性数据集来源于Kaggle，包含358条记录和

Python实现保险公司的理赔支出预测源码+数据集，健康保险理赔预测源码，对特征数据进行可视化分析，用线性回归模型进行机器学习

国家中小学智慧教育平台（课件、课本、视频）下载器