Loan Process Automation

The Project has three main parts, utilising two Bots. The first Bot- “LoanFormUpload” is responsible for uploading the Loan Forms (pdf forms) to the IQ Bot, which would then go through each form and create a csv file with the field name and details. The second Bot - “LoanFormDownload” will then download the csv files stored in the Automation Room Cloud to the host system in the specified folder. The last step is the execution of a Python script, which would run each file through the trained model, predict if the Loan should be approved and store all final values in another csv file in the specified directory. The dataset used to train the AI model which would predict if the Bank should approve the loan is a publicly available Kaggle Dataset. It consists of 13 features and over 1000 entries in total. After feature engineering, 11 features were chosen for the final model. Missing values were filled using metrics such as mathematical average, maximum and minimum frequency or were randomised. A script was created which would take the csv files downloaded by the Automation Anywhere Bot, perform the necessary data cleaning and normalizations and then predict if the Loan should be approved.