A Novel Deep Belief Network Based Approach for Retail Store Sales Prediction During Peak Demand Seasons and its Performance Comparison over K-Nearest Neighbour Technique

Aim: The research is about the Novel Deep Belief Network (NDBN) approach for Retail Store Sales Prediction during peak demand seasons and its performance comparison over K-Nearest Neighbour Technique (KNN). Materials and Methods: Deep Belief Network (N=10) and K-Nearest Neighbour algorithm (N=10) samples were considered based on the clinc calc online sample size calculator for predicting the accidents that happened in terms of accuracy. Two sample groups are taken into consideration and tested, G-power is the calculation that contains two different groups, alpha (0.05), power (80%), and environment ratio. Results: The Novel Deep Belief Network algorithm achieved 84.53% accuracy and K-Nearest Neighbour has 74.24%. This NDBN appears to have significance of p equal to 0.02 for the K-Nearest Neighbour, that is p less than 0.05 using independent sample T-test analysis. From the result, it proves that the Deep Belief Network approaches predict the retail sales store prediction.