leanding_loan_club_defaulters_prediction

1. Objective: Task is to determine if person is defaulter or not => Classification problem.

1.1 Data overview:

Number of data points: 42,538
Attributes: 115

1.2 Type of Machine learning problem

Task is to predict views of TED talk videos => Regression problem

1.3 Performance Metrics used

1. Accuracy
2. Confusion Matrix
3. Prcision , Recall & F1 Score

2. Exploratory Data Analysis

2.1 Distribution of target variable

- Observation: Imbalanced Data

2.1 Bivariate analysis on Interest Rate V/S Target variable (Defaulters)

- Observation: Even if we have imbalanced data people with higher interest rate tends to fall under defaulter category

2.1 Bivariate analysis on Interest Rate V/S Month Term

- Observation: People with higher Interest rate tends to have 60 months term instead of 36

2.3 Bivariate analysis on Interest Rate V/S Grades (views)

- Interest rate between 5-9 falls under A grade - Interest rate between 10-12 falls under B grade - Interest rate between 13-16 falls under C & D grade - Interest rate greater than 17 has the risk of comming under Defaulters category i.e. Grade F&G

3. Feature Engineering

• Unecessary variables which contains zip , id or only single category has been dropped
• Features like Grade, Interst rate in %, sub_grade which holds higher information are been handled with Ordinal encoding.

4. Feature Selection

4.1 f_regression to get feature importance, Dropped features with higher P-value with threshold > 0.3

5. Comparison of Accuracies on diffrent Machine learning Models used.

1. Gaussion Naive Bayes

2. Logistic Regression

3. Decision Trees

4. Random Forest Classifier

Conclusion & Buisness values

So after loading the data we started with EDA process to understand the data through diffrent types of Univariate,Bivariate & Multivaiate tools and also handles outliers and NaN values. In feature engineering we have created some of the features as well as removed some unwanted features which added less value . Features such as Grade, Sub_Grade, Intreset rate, term etc etc played major roles to understand wheather the person is defaulter or not. Finally we have compared all the models w.r.t.o their Acc, Confusion Matrix , Precison & Recall and all the models have performed better in this case.