Home / Current Issue / Paper 1717475
Customer Lifetime Value Prediction for Fastrack Watches Using Machine Learning Regression
Subject area: Science,Engineering and Technology · Area of research: Business Analytics
DOI: https://doi.org/10.64388/IREV9I11-1717475
Abstract
This study evaluates whether Customer Lifetime Value (CLV) for Fastrack watch buyers can be reliably predicted from survey-based features using supervised regression models in India. Primary data was collected from 106 respondents through a structured questionnaire. The study assessed demographic profiles, purchase behaviour, satisfaction levels, and brand loyalty of Fastrack customers. A composite CLV score was engineered from average order value, purchase frequency, and customer lifespan proxies. Six regression models Linear Regression, Ridge Regression, Lasso Regression, ElasticNet, Random Forest Regressor, and Gradient Boosting Regressor were trained and evaluated using MAE, RMSE, MAPE, R², and 5-fold cross-validation. Key findings reveal that Linear Regression achieved the highest test-set R² of 0.8476, while Gradient Boosting achieved the lowest MAPE of 11.56% and the best cross-validated R² of 0.7897. Price Range (28.7%), Purchase Frequency (19.8%), and Usage Duration (15.6%) emerged as the top three predictors of CLV, collectively accounting for 64.1% of total feature importance. Demographic features (Gender, Location, Age Group) ranked lowest in importance. The study concludes that CLV can be reliably predicted from survey data (R² > 0.84), and Gradient Boosting is recommended for real-time CLV scoring deployment.
Keywords
Customer Lifetime Value, CLV Prediction, Machine Learning Regression, Fastrack Watches, Gradient Boosting, Feature Importance.
How to cite this paper
@article{1717475,
author = {Dr. M. Iswarya, C. Logesh},
title = {Customer Lifetime Value Prediction for Fastrack Watches Using Machine Learning Regression},
journal = {Iconic Research And Engineering Journals},
year = {2026},
volume = {9},
number = {11},
pages = {720-728},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1717475.pdf},
abstract = {This study evaluates whether Customer Lifetime Value (CLV) for Fastrack watch buyers can be reliably predicted from survey-based features using supervised regression models in India. Primary data was collected from 106 respondents through a structured questionnaire. The study assessed demographic profiles, purchase behaviour, satisfaction levels, and brand loyalty of Fastrack customers. A composite CLV score was engineered from average order value, purchase frequency, and customer lifespan proxies. Six regression models Linear Regression, Ridge Regression, Lasso Regression, ElasticNet, Random Forest Regressor, and Gradient Boosting Regressor were trained and evaluated using MAE, RMSE, MAPE, R², and 5-fold cross-validation. Key findings reveal that Linear Regression achieved the highest test-set R² of 0.8476, while Gradient Boosting achieved the lowest MAPE of 11.56% and the best cross-validated R² of 0.7897. Price Range (28.7%), Purchase Frequency (19.8%), and Usage Duration (15.6%) emerged as the top three predictors of CLV, collectively accounting for 64.1% of total feature importance. Demographic features (Gender, Location, Age Group) ranked lowest in importance. The study concludes that CLV can be reliably predicted from survey data (R² > 0.84), and Gradient Boosting is recommended for real-time CLV scoring deployment.},
keywords = {Customer Lifetime Value, CLV Prediction, Machine Learning Regression, Fastrack Watches, Gradient Boosting, Feature Importance.},
month = {May},
doi = {https://doi.org/10.64388/IREV9I11-1717475}
}