The present study was conducted on seven tourism companies in BIST Tourism Index in order to describe continuous financial factors which affect net profit margin (NPM) as a continuous response variable through CART (Classification and Regression Tree), CHAID (Chi-Square Automatic Interaction Detector), Exhaustive CHAID and MARS (Multivariate Adaptive Regression Splines) algorithms. In the present study, the data of these companies from the period 2011-2017 were evaluated. Predictive performances of CART, CHAID, Exhaustive CHAID and MARS in predicting NPM were measured based on model goodness of fit criteria, viz. r (Pearson correlation coefficient between actual and predicted values in NPM), coefficient of determination (R2), adjusted coefficient of determination (Adj. R2), standard deviation ratio (SDRATIO), root of mean square error (RMSE), global relative approximation error (RAE), mean absolute deviation (MAD), Akaike’s information criterion (AIC) and the corrected Akaike’s information criterion (AICc). In the study, financial factors used in the prediction of NPM were current ratio (CR), acid-test ratio (ACTR), asset turnover ratio (ASTR), accounts receivable turnover ratio (ACRTR), equity turnover ratio (EQTR), short term liabilities to total assets ratio (SHTLTAR), long term liabilities to total assets ratio (LOTLTAR), total assets to equity ratio (TOAER), long term liabilities to equity ratio (LOLER) and total debt to total assets ratio (TODTAR) as predictors. In the prediction of the NPM and the description of the influential financial factors influencing the NPM, the highest predictive accuracy was obtained by MARS algorithm (r= 0.980) and the statistically significant order was found as MARS (r= 0.980)> Exhaustive CHAID (r= 0.915)= CART (r= 0.873)= CHAID (r= 0.868) algorithms.

In conclusion, the achieved results indicated that, i) the regression tree diagram constructed by Exhaustive CHAID algorithm displayed that tourism companies with LOTLTAR< 0.3715 and EQTR< 0.0311 had the highest average NPM of 2.778, ii) CART tree-based algorithm showed that the companies with EQTR>-0.2125 and ASTR< 0.0246 had the highest average NPM of 4.226, iii) the diagram of CHAID tree-based algorithm revealed that the companies with TODTAR< 0.6145 and EQTR< 0.0311 had the highest NPM with the average of 2.778. It is recommendable that data mining algorithms capture optimal cut-off values of influential factors, which may ensure the highest NPM values.