Open AccessMisclassified Reduced Instance and Stochastic Gradient Descent with Logistic Regression Model for Customer Churn Prediction
Author(s) -
Isabella Amali*,
Rahul Kumar,
Rakesh Mohan
Publication year - 2020
Publication title -
international journal of innovative technology and exploring engineering
Language(s) - English
Resource type - Journals
ISSN - 2278-3075
DOI - 10.35940/ijitee.d1784.029420
Subject(s) - computer science , logistic regression , classifier (uml) , benchmark (surveying) , boosting (machine learning) , artificial intelligence , gradient boosting , machine learning , stochastic gradient descent , regression , deep learning , predictive modelling , data mining , random forest , artificial neural network , statistics , mathematics , geography , geodesy
Customer Churn Prediction (CCP) is a difficult problem found to be helpful to make decisions due to the rapid growth in the number of telecom providers. At present, deep learning models are familiar because of the significant improvement in different areas. In this paper, a deep learning based CCP is introduced by the use of Stochastic Gradient Boosting (SGD) with Logistic regression (LR) classifier model. By the integration of SGD and LR, effective classification can be accomplished. To further improve the classifier efficiency, misclassified instances are removed from the dataset. Then, the processed data is again provided as input to the classification model. The presented SGD-LR model is validated on a benchmark dataset and the results are examiner with respect to different measures. The experimental outcome pointed out the projected model is superior to available CCP models on the identical dataset.