Aim of the Project : To predict, analyze and provide the counter measures to prevent the customers’ from Churning. Build a Model to predict the Customers’ likely to be churned using Data Science, and Machine Learning Applications.

Business Understanding : The data used to build a Customer Churn Prediction Model, consists of data related Telecom domain, and their behavioral trends w.r.t. Churning across 4 years of time span.

Life Cycle of the Project: Extracted the data from Kaggle open source. Performed Data Exploration to understand the descriptive stats w.r.t. the data, in order to manipulate the data to build a robust ML model. Used Pandas, matplotlib & Numpy for Data Pre-processing, to decrease the redundancy, by taking care of the missing values, and duplicates. Performed Data Exploration based on UNIVARIATE ANALYSIS & BIVARIATE ANALYSIS. Carried out extensive EDA using KDE-PLOT, BARPLOT, & HEATMAPs to understand the data ( Monthly || Total Charges), its outliers and as its a classification problem. Trained the model using Sci-Kit Learn. Used Random Forest Classifier to build the ML Model with Utmost Accuracy. Used SMOTE + ENN technique, to convert data to 50:50 ratio & Enhance model accuracy.

Results from the Project Check out the Detail Project Overview on GitHub Repository

kaggle notebook

• scene