DATA QUALITY AWARE CUSTOMER RETENTION PREDICTION

Customer retention has become an important focus for many businesses, as keeping existing
customers is usually more cost-effective than attracting new ones. However, predicting whether a
customer will stay or leave a service can be difficult when the available data contains quality issues.
Problems such as missing values, duplicate records, inconsistent data formats, and noisy information can reduce the accuracy of machine learning models. Therefore, improving data quality before
building predictive models is an essential step.
This project presents a Data Quality-Aware Customer Retention Prediction System that
combines data quality analysis with machine learning techniques. The system first analyzes the
dataset to identify common data quality problems such as missing values, duplicates, outliers, and
inconsistencies. After detecting these issues, different preprocessing techniques such as data cleaning,
normalization, and feature engineering are applied to improve the quality of the dataset. Once the data
quality is enhanced, machine learning models are trained to predict customer retention more
effectively. The project also compares model performance before and after data quality improvement
to highlight the impact of clean and reliable data on prediction accuracy. Different algorithms such as
Logistic Regression, Random Forest, and Gradient Boosting are used to determine which model
performs best for customer retention prediction.