A stochastic logistic sigmoid regression using convex programming and clustering

Abstract

Logistic regression is one of the regression analysis methods that was studied a long time ago and its applications are widely used in many classification tasks. In this paper, a stochastic model is proposed by our that calls stochastic logistic sigmoid regression. This problem is solved by the new approach that transforms a deterministic problem into a stochastic problem and solves it by a convex programming problem. Besides, to estimate the mean and variance-covariance matrix of random variables, clustering algorithms, and quantile estimation are applied. The effectiveness of the model is evaluated by metrics for evaluating the performance of logistic regression. The results of the proposed algorithms, which are overcome over 1 to 2 percent with an accuracy score on three datasets, include many different fields data. They are also better than the ordinary logistic regression model on the same dataset with evaluation metrics, examples: f1 score, precision score, recall score, confusion matrix, et cetera.