Exploratory Data Analysis Bank Fraud Case Study
- 1. BANK LOAN EDA CASE STUDY PREPARED BY: MR.ABHISHEK LAL MS.LUMBINI SARDARE
- 2. PROBLEM STATEMENT ● The loan providing companies find it hard to give loans to the people due to their insufficient or non-existent credit history. Because of that, some consumers use it as their advantage by becoming a defaulter. ● This case study aims to identify patterns which indicate if a client has difficulty paying their installments which may be used for taking actions such as denying the loan, reducing the amount of loan, lending (to risky applicants) at a higher interest rate, etc. This will ensure that the consumers capable of repaying the loan are not rejected. Identification of such applicants using EDA is the aim of this case study. 2
- 3. ANALYSIS DONE Steps ● Data Understanding & preparation ● Data cleaning & Manipulation ● Data Analysis- univariate Bivariate ● Presentation & recommendations 3
- 4. Data Understanding and preperation Data Understanding and preparation Step1:Data sourcing a:Imported the CSV files and then read the data B:Data Inspection: we checked the shape of the data ,as well as checked data,checked the data type of the given data, we Inspected the dataframe for dimensions, null-values, and summary of different numeric columns. C:Inspecting Null Values in rows which are greater than 50%, -Checked columns with missing values & dropped colums with missing values greater than 50% -impute values for columns with missing values of aprox. 13% and then uniques values are less and these columns give day, week, month, year etc we can replace null values with 0 -Dropped unwanted columns -changed ID to Object & checked datatypes of columns -converted negative values to positive -divided income into credit groups -variables indicate number of days, hrs, months, etc. we can conclude that these are columns with categorical values.And regarding the null values, as it can be seen from the data above, since these are categorical columns, and there mode value is 0 for all these variables. so, it is safe to impute the null values with 0. 4
- 5. DATA CLEANING & MANIPULATION we have the final data frame after dropping unwanted columns, we now check for data types of each columns and change their datatype based on the values they contain
- 6. Checking for data quality issues and binning of continuos variable: -checked for all data types and they they are converted into appropriate standard format -Performed datatype conversion & changed negative to positive ,converted data into psotive format,and rounded of required values 6
- 7. 7 Found outliers using boxplot
- 8. Changed the credit amount,checked imbalance % and Ploted Distribution of Target values where credit amount and income amount imbalance is shown 8
- 9. 9 HEAT MAP FOR CORELATION MATRIX
- 10. DATA ANALYSIS Data Imbalance check Categorized target =0 and Target=1 Univariate analysis For defaulter & non defaulter using SNS Boxplot. Defaulters are within the age group of 31-50 years 10
- 11. 11
- 12. 1.CUSTOMER WITH SECONDARY & HIGHER SECONDARY AS WELL AS LABOURERS ARE HIGHEST IN DEFAULTERS 12
- 13. CUSTOMERS LIVING IN HOUSES & APARTMENTS ARE HIGHEST IN DEFAULTERS 13
- 14. BIVARIATE ANALYSIS Income for defaulters is very low irrespective of no.of years employed 14
- 15. SCATTER PLOT NON DEFAULTER GET MORE CREDIT FOR THEIR ANNUITIES 15
- 16. 16 RANGE OF CUSTOMERS WITH HIGHER DEGREE IS HIGHER IN NON-DEFAULTERS
- 17. 17
- 18. 18
- 19. 19
- 20. 20
- 21. RECOMMENDATIONS & CONCLUTIONS ● Range of customer with higher degree is higher in non defaulters ● Non defaulter get more credit for their annuities ● Income for defaulter is very low irrespective of no.of years employed ● Customer living in housing/Apartments are highest category in defaulter ● Labourer's are biggest category in defaulters ● Customer with secondary/secondary special are highest in both defaulters and non defaulters ● Purpose of finding defaulters and their details, steps involved was data cleaning, data analysing,drawing inferences from it and the graphs for the same has been attached 21
- 22. 22 THANKS!