Machine Learning For Predictive Financial Analytics By Applying Naïve Bayes Algorithm

Abstract

Naïve Bayes is an important tool for machine classifier. It classifies the data sets by maximum posterior
decision rule. In the present research study naive bayes is applied to get the probabilistic analysis. The
naive bayes predicts on the basis of prior knowledge and current evidence. The variables used are
financial ratios 1.Return on asset (ROA) is measures of Profitability 2.Debtors turnover ratio (DTO)
Inventory turnover ratio (ITO) as measures of efficiency 3.Debt to equity ratio(DER) measures of long
term solvency 4.Current ratio as measures of liquidity. The ratios are defined by using class satisfactory
(within Range) and dissatisfactory (not within Range). The study is design with five step model. Decide
the range for converting financial ratios into binary as satisfactory (within Range=1) and dissatisfactory
(not within Range=0), Create frequency table for class, Calculate prior probability, Create likelihood
table , Calculate Posterior probability and Analysis by using Confusion Matrix . The accuracy of model is
0.45