Authors:
S. Silvia Priscila, S. Suman Rajest, Sai Nitisha Tadiboina, R. Regin, Szeberényi András 

Addresses:
1Department of Computer Science, Bharath Institute of Higher Education and Research, Tamil Nadu, India. 2Department of Research and Development (R&D), Dhaanish Ahmed College of Engineering, Chennai, Tamil Nadu, India. 3Department of Information Technology, Geico, Maryland, USA. 4Department of Computer Science and Engineering, SRM Institute of Science and Technology, Chennai, Tamil Nadu, India. 5Institute of Marketing, Budapest Metropolitan University, Budapest, Hungary. silviaprisila.cbcs.cs@bharathuniv.ac.in1, sumanrajest414@gmail.com2, stadiboina@geico.com3, regin12006@yahoo.co.in4, aszeberenyi@metropolitan.hu5

Abstract:

Today, a group of supermarkets requires a consistent ridge of their yearly sales. This primarily results from a need for knowledge, resources, and the capability to estimate sales. Conventional statistical methods for supermarket sales are important and often lead to predictive models. In the age of big data and powerful computers, machine learning is the standard for sales forecasting. This comprehensive literature review examines superstore sales prediction models using ML and DL. This article review focuses on superstore sales prediction using machine learning and deep learning in data mining. Finally, DL is the best SSP for results. DL models market movements well. Automatic feature extraction models and forecasting strategies have been tested with various inputs. DL algorithms process large real-time datasets better. DL research found the best hybrid processing methods for real-time stock market data. DL and ML methods predict the client's response and identify its factors. DL and ML algorithms are evaluated using Rodolfo Saladanha marketing campaign data. Four metrics precision, recall, F-measure, and accuracy compare ML and DL algorithms. MATLAB tested these methods. LSTM, CNN, LR, RF, and LR algorithms were used to compare results to well-known ML and DL algorithms. Artificial Convolutional Neural Network (ACNN) is compared to RF, LR, CNN, and LSTM. The proposed superstore sales prediction algorithm outperformed the others. The proposed model predicted superstore sales with a validation accuracy of 93.90 percent, outperforming current and suitable baselines.

Keywords: Deep Learning (DL); Superstore Sales Prediction (SSP); Random Forest (RF); Linear Regression (LR); Convolutional Neural Network (CNN); Long Short-Term Memory Networks (LSTM); Deep Neural Network (DNN).

Received on: 06/10/2022, Revised on: 22/11/2022, Accepted on: 03/01/2023, Published on: 05/02/2023

FMDB Transactions on Sustainable Computer Letters, 2023 Vol. 1 No. 1, Pages: 1-11

• Views : 1204
• Downloads : 24
• Cited by: 88

Download PDF