Sales Forecasting Models
This document summarizes the process of training and testing machine learning models for forecasting sales projects
The dataset, includes 12,330 sessions, reveals that 84.5% (10,422 sessions) were categorized as negative class samples, indicating no purchase was made. Conversely, 15.5% (1,908 sessions) were categorized as positive class samples, ending with a shopping event.
Using Scikit-Learn for Model Development
Scikit-Learn is a powerful open-source library in Python for building and deploying predictive models. It provides a wide range of algorithms and tools for data preprocessing, model selection, and hyperparameter tuning - key steps in developing accurate sales forecasting models.
Results
Based on the metrics, SVM_Model_3 with an RBF kernel and standardized data appears to be the strongest performer, achieving the highest F1 score and AUC while maintaining excellent overall accuracy.