1.Checking data types of the features and any feature contain NAN values or not. 2.Descriptive statistics include those that summarize the central tendency, dispersion and shape of a dataset’s distribution 3.Checking for multicolinearity 4.Checking the ditribution of the Features. 5.Cheking if the features have any outlier or not. 6.Checking the frequency of the target varibale. If it's a imbalanced calssification or not.
- Remove the highly correleated features.
- Remove the outlier from the features
- Done scaling to the features
- Applied Oversampling technique to solve the imbalanced class problem.
- Split the the dataset with ratio of 4:1
- Fit the Random Forest Model
- Make a list of top 10 important feature using Feature Importance
- Select top 10 most important from the train_dataset
- Done Preprocessing on the selected dataset
- Split the dataset with 4:1 ratio
- Train the train-data using Random Forest
Model evaluation has been done by some of the most important metics-
- ROC_AUC Score
- Confusion Matrix 3.Precision
- Recall
- F1-Score
- Tried different paramater using Grid Search to check if it's possible to increase the model performance or not.