Churn Prediction Project

Overview

This project focuses on predicting customer churn — determining whether a customer will leave a company’s service or not.
Such predictive modeling helps businesses (especially in telecom, banking, and SaaS industries) to identify at-risk customers and take early retention actions.

---

Dataset

• Name: Telco Customer Churn Dataset
• Source: Kaggle Dataset
• Size: 7032 records and 21 features
• Content: Customer information, service usage, contract types, payment details, and churn status.

---

Project Workflow

Data Preparation

• Inspected and handled missing values
• Removed irrelevant columns (customerID)
• Converted categorical variables using LabelEncoder
• Scaled numerical features as needed

Exploratory Data Analysis (EDA)

• Visualized distributions (countplot, boxplot, violinplot)
• Checked correlations between variables (corr())
• Explored patterns between churn and service types

Model Training

The following classification models were built and evaluated:

• Logistic Regression
• Support Vector Machine (SVM)
• Decision Tree
• Random Forest
• AdaBoost
• Gradient Boosting

Hyperparameter tuning was applied using GridSearchCV for Decision Tree (as an example). For the rest of the models, the default model mode of the models themselves was used.

Model Evaluation

The models were compared using multiple performance metrics:

For numerical analyses

• classification_report which includes:
  -- Accuracy -- Precision -- Recall -- F1-score

To analyze numbers in the form of a Display chart

• ConfusionMatrixDisplay

A summary table and visual comparison were created to display the results clearly.

---

Results

The Gradient Boosting and Adaboost classifiers achieved the highest accuracy. Among them, Adaboost achieved an accuracy of approximately 81% on the test set.

Top influential features:

• Contract Type
• TechSupport Availability
• Customer Tenure

---

Libraries Used

• pandas
• matplotlib
• seaborn
• scikit-learn
• joblib
• warnings

---

How to Run

Clone the repository:

git clone github cd Churn_Prediction

Install dependencies:

 pip install -r requirements.txt 

Open the Jupyter notebook:

jupyter notebook notebooks/Churn_Prediction.ipynb

Run all cells to train and evaluate the model.

---

Author ✍️

Author: Ali
Field: Data Science & Machine Learning Student
Email: ali.hz87980@gmail.com
GitHub: ali-119