The central idea of this project is to develop a web application that collects input from the user and can show the probability of churn for the data given by the user as an input. This project has been done with the motive of developing the same for Indian banking organizations, with which any banking firm can predict the churn of its customers with ease. Since there is no proper past data regarding the customers in India acquired, this project takes data from the banks located in countries like Germany, Spain, and France, and the same idea can be implemented in the future for Indian banking organizations. Behind this application, there has been a trained neural network model deployed, with which the application is capable of delivering accurate results. Moreover, the application is designed in such a way to display an interactive dashboard for the user, from which he/she can infer insights from the visualizations and can compare themselves with the past data. To train a machine learning model to predict the churn of the customer by taking several features into consideration. • To prepare the data for analysis – it includes data pre-processing, data cleaning etc. • To provide insights on the data through Exploratory Data Analysis. • To create a Web Application, which has a trained model in its back end and could deliver results on the data given by a user
-
DATA PART:
- DATA IMPORTING
- DATA PREPROCESSING & CLEANING
- EXPLORATORY DATA ANALYSIS
- FEATURE TRANSFORMATION AND ENCODING
-
VISUALIZATIONS:
- FINDING RELATIONS BETWEEN FEATURES
- ANAMOLY DETECTION
-
MODEL BUILDING:
- TRAIN TEST SPLIT
- CREATING A NEURAL NETWORK USING SEQUENTIAL API FROM KERAS
- TRAINING AND TESTING
-
VALIDATION:
- CONFUSION MATRIX, ACCURACY
-
MODEL DEPLOYMENT:
- SAVING THE MODEL
- SETTING UP ALL FILES
- CREATING A FRONT END USING HTML AND CSS
- USING FLASK TO CREATE A WEB APPLICATION