DEVELOPMENT OF TIME-SERIES-BASED MLOPS ARCHITECTURE FOR PREDICTING SALES QUANTITY IN MICRO, SMALL, AND MEDIUM ENTERPRISES (MSMES)

Abstract

Micro, Small, and Medium Enterprises (MSMEs) constitute a significant portion of the economy in many developing countries, playing a vital role in employment generation and economic growth. Sales performance is a critical factor for MSMEs, influenced by various internal and external factors. Time-series analysis offers a valuable tool to predict sales quantities by analyzing historical data and identifying patterns and trends. In this context, the SARIMAX (Seasonal Autoregressive Integrated Moving Average with Exogenous Variables) model emerges as a suitable method to forecast future sales, leveraging both historical data and external variables. This research explores the synergy between time-series analysis, specifically SARIMAX modeling, and MLOps (Machine Learning Operations). Finally, this research aims to provide a framework for the practical application of MLOps to enhance sales forecasting and decision-making processes within MSMEs, fostering their growth and sustainability in a competitive market landscape.