sonusinghmehra090/CreditWise_LoanApprovalSystem_-MinorProject- ? reverse-engineered prompt

Build me a simple end to end loan approval prediction project in a Jupyter notebook.

I want it to load the provided loan dataset, explore the data a bit, clean missing values, do the needed encoding and scaling, and then train and compare Logistic Regression, Decision Tree, and Random Forest models for binary loan approval prediction.

Please include basic feature engineering and feature selection if it helps, and make sure the workflow avoids data leakage by using proper train and test splitting and a pipeline where it makes sense. I’d also like model tuning with grid search, then a clear evaluation using accuracy, precision, recall, and f1 score so I can see which model performs best.

At the end, show the results in a simple way and save or display a chart of model accuracy like the one in the repo. Keep the notebook clean and easy to follow, and use pandas, numpy, and scikit learn.