Abstract:

It is seen that many of the campaigns implemented by companies providing online shopping services to increase their product sales and acquire new customers have not been successful. In order to increase product sales in online shopping and to make campaigns successful, various data of 394 users who shopped on the online market were collected through a survey. In the study, a model that provides prediction of the shopping tendencies of online market users from the personal care category has been presented by applying frequently used machine learning algorithms. Decision Trees, K-Nearest Neighbor, Gradient Boosted Trees, Random Forest and Logistic Regression methods were used for modeling. In addition, analysis has been done with AUC (Area under the Curve), recall, f1-score values to compare the performance of the applied models. As a result of the study, the highest performance was obtained with the random forest method with an accuracy rate of 0.928 and AUC of 0.92. The model which applied with Gradient Boosted Trees method was achieved the lowest performance values with an accuracy rate of 0.704 and an AUC value of 0.70. According to the findings in the study, it has been analyzed that users whose properties are "under the age of 43-47, high daily internet user, do not choice to pay by credit card at the door" prefer shopping from the personal care category. When the results obtained are examined, it is predicted that the model set forth in the study will enable companies to determine their target customer groups better, and thus, unnecessary investments can be prevented.

Keywords: