Machine Learning Based Inventory Management for Medical Stores: Predicting Seasonal and Regular Drug Demand

Nowadays medical supply chain is important to ensure that medicines are delivered to people on time satisfying the requirement based on seasonality and location. However, issues like stockouts, overstocking, and waste hinder efficient operations especially when the changes in the seasons and consumption trends are not well addressed. At present, most systems employ mostly reactive strategies compared to predictive analysis which leads to sub-optimal inventory management. This study attempts to solve these challenges by applying machine learning (ML) methods to the data analysis and demand forecasting. This study received sales data from several medical retailers over six months to group drugs as seasonal or regular and then makes predictions for future demand. The main aim is to decrease waste, increase the availability of products and improve the general inventory management. The dataset features elements such as store ID, medicine name, Quantity_sold, date, and type which are either seasonal or regular. This paper applies Exploratory Data Analysis to determine sales trends, to identify the top-selling drugs, and to analyze the temporal patterns of demand. Random Forest Regression and Logistic Regression were employed in the study with different assessments for seasonal and routine drugs. Summary statistics and trend analysis were also used in the research. The results of the summary statistics showed that the sales of the seasonal products are more unpredictable than the non-seasonal products which mean that special forecasting techniques should be used for creating a data-driven framework for the optimal management of medical store inventory. Future work could also involve combining geospatial and socioeconomic factors in the model to enhance the prediction accuracy as well as using advanced deep learning algorithms.