Neural Networks for Customer Classification Through Clickstream Analysis

Business intelligence (BI) refers to the strategies, technologies, and tools used by organizations to analyze and interpret large amounts of data, with the goal of gaining actionable insights and making informed business decisions. BI can incorporate predictive analytics techniques to forecast future trends and outcomes. By analyzing historical data and applying statistical models, enterprises can make predictions about customer behavior, market trends, and other factors that impact their business. The research aims to classify customers based on their clickstream usage patterns by utilizing artificial neural networks (ANNs). For this purpose, a clickstream-based database collected from the IMOLKO company’s website via the Google Analytics platform was employed. Several experiments were conducted using Monte Carlo cross-validation (MCCV) to adjust the number of hidden layers in the ANNs and utilize different proportions of testing data. To evaluate the model’s performance the accuracy, sensitivity, specificity, positive predictive value, negative predictive value and the F1 score were calculated. The evaluation metrics, calculated through MCCV, exhibited low standard deviations, indicating that the ANN classifier is robust and not significantly affected by random variations in the testing-train database split. The ANN exhibited a sensitivity, specificity, positive predictive value, negative predictive value, and F1 score above 80%. However, it is important to note that there is a 9% of probability of false positives, which slightly affects the F1 score and sensitivity. In conclusion, the ANN employed in this study demonstrated their effectiveness as a classification technique for predicting customers of the corporation IMOLKO C.A.