Linear Regression and Artificial Neural Networks based Efficient Sales Forecasting Model with Increased Prediction Accuracy
S. Ravi Teja Reddy, P. Malathi · 2022 4th International Conference on Inventive Research in Computing Applications (ICIRCA) · 2022
The study's goal is to uncover the key factors that influence sales, as well as conduct an experiment to determine the optimal algorithm among Linear Regression and innovative Artificial Neural Networks. Materials and Methods: Description Predicting a product's future pricing using a linear regression technique (N=10) and a novel ANN approach (N=10) with supervised learning that incorporates past data. The sample size per group is 80, and G power is 80%. Results: A statistically significant difference between Linear Regression and new Artificial Neural Networks was identified, with p = 0.000 (0.05,2-tailed). Between the groups, the average statistical significance was calculated. The new ANN has a 95% accuracy rating, compared to 84 percent for the linear regression classifier. Conclusion: The results proved that the novel artificial neural networks algorithm is significantly better for sales forecasting than the linear regression algorithm within the study limits.