Sales Forecasting Based on Time Series Analysis

• Published in: 2024 International Conference on Science Technology Engineering and Management (ICSTEM) pp. 1 - 7
• Main Authors: S, Selvakumar, G, Renugadevi, N, Vinishah, R, Yashwanth
• Format: Conference Proceeding
• Language: English
• Published: IEEE 26.04.2024
• Subjects: Accuracy; Machine Learning; Predictive models; Profitability; Sarimax; Single-variable Prediction; Stationarity; Supply chain management; Temporal Analysis; Time series analysis; Transforms; Weather forecasting
• DOI: 10.1109/ICSTEM61137.2024.10560659

Effective demand forecasting and inventory management are essential elements in today's business environment, ensuring optimum inventory levels and cost minimization. This study presents an alternative solution to overcome these challenges, using a seasonal autoregressive integrated moving average (SARIMAX) model with exogenous intercepts. The research explores how implementing the SARIMAX model can streamline supply-demand forecasting and enhance inventory management procedures. By combining historical sales data with relevant external factors, the SARIMAX system increases accuracy in anticipating requirements. This accuracy provides decision makers with valuable insights to strategically plan production, distribution, and replenishment efforts. The results of this study highlight the transformative potential of the SARIMAX model to transform supply chain management practices. By leveraging data-driven insights, companies can proactively drive market transformation, optimize distribution, and increase profitability. This paper contributes to the expanding body of knowledge about demand theory and inventory development, and provides important insights for practitioners and researchers. Temporal analysis methods serve as the cornerstone for examining the evolution of processes or metrics over time. Machine learning stands as an artificial intelligence approach enabling systems to autonomously learn and enhance their performance through accumulated experience. Analyzing time series involves employing various techniques to discern meaningful statistics and other characteristics from the data. The analysis typically proceeds through five stages: data acquisition and preparation, feature manipulation, model evaluation, model selection, and visualization. Time series models find diverse applications such as sales prediction and weather forecasting. This research aims to elucidate univariate forecasting, the stability concept, autoregressive integrated moving average (ARIMA) models, and seasonal ARIMA models with exogenous factors. Effective management of a company's sales is crucial for future profitability and minimizing losses. Therefore, stakeholders must diligently manage the company's finances to optimize profits and mitigate risks.