AI-driven forecasting and stock classification for pharmacy inventory

Overview

Developed and evaluated machine-learning models to support pharmacy medication inventory management, focusing on forecasting and stock-level classification to help identify potential low, normal, or high inventory conditions.

My Role

Data Preparation • Machine Learning • Model Evaluation • Analysis

I prepared the dataset, selected relevant features, trained multiple machine-learning models, compared their performance, and analysed which approaches were most suitable for pharmacy inventory decision support.

The Challenge

Pharmacy inventory needs to balance product availability with the risk of overstocking or shortages. The objective was to explore whether machine learning could help classify medication stock levels and support more informed inventory planning.

The Solution

The prototype used structured pharmacy sales and inventory data to:

• Prepare and clean the dataset
• Engineer/select relevant variables
• Train multiple classification models
• Compare model performance
• Analyse stock-level predictions
• Identify the most useful variables for inventory decisions

Key Features Considered

Stock Level • Month • Year • COVID Flag • Unit Price

Outcome

The project demonstrated how different machine-learning algorithms can be compared for pharmacy medication stock-level classification and inventory forecasting, supporting a more data-driven approach to inventory management.

Visuals

1. Confusion matrix

2. Feature importance chart

3. Pharmacy Inventory ML Pipeline