A Fine Tuned-based Framework to Predict Salesforce Data using Machine Learning in Business Analytics

Abstract

Sales forecasting is one of the critical areas in business analytics where business organizations aim to enhance efficiency and, therefore, revenues. An excellent example of a CRM program is Salesforce, which produces massive amounts of sales data that are essential for forecasting and decision-making. Data analysis involves the use of complex and effective tools for its processing. This study proposes a framework based on the following classification algorithms: Support Vector Machines (SVM), Decision Trees (DT), and Random Forests (RF). The proposed framework follows a fine-tuned approach to improve the prediction of sales data. Regarding the fine-tuning of these algorithms, it was observed that specific changes were required within the hyperparameters to better relate to the inherent patterns and other factors that exist in the sales data. The optimization process was very crucial in improving the performance of the model. The proposed framework was used on a sales dataset and evaluated in terms of accuracy, precision, data loss, and F1 score. Fine-tuned algorithms had higher accuracy and lower data loss.