The project focuses on using Zero-One Integer Programming to optimize the selection of skincare products for an upcoming Boxing Day Sale. The goal is to maximize revenue while catering to customer needs based on skin types and budget constraints.
- Optimize skincare product mix: Select the best combination of skincare products tailored to different skin types and budget categories.
- Maximize revenue: Use integer programming to determine the optimal selection of products to maximize sales revenue.
- Handle real-world constraints: Include constraints such as non-repeating products across different price categories and skin types.
Decision_Making.ipynb: Jupyter notebook containing the Python code used to build and solve the integer programming model using Gurobi.Decision Making_report.pdf: The individual report explaining the problem formulation, model, results, and recommendations based on the analysis.Skincare_Product_Data.xlsx: The dataset of skincare products used for the analysis, including product types and prices. The entire dataset is from Kaggle website: Skincare Recommendation Engine.
To run the notebook (Decision_Making.ipynb), ensure you have the following installed:
- Python 3.13.0
- Libraries:
pandas,gurobipy,numpy
- Install the necessary libraries using:
pip install pandas gurobipy numpy
- Open the Jupyter notebook in your environment and run the cells sequentially to replicate the analysis and optimization results.
- The notebook includes:
- Data preprocessing: Cleaning and structuring the skincare product data.
- Model building: Formulating the objective function and constraints.
- Optimization: Solving the Zero-One Integer Programming model using Gurobi to find the optimal product mix.
- Optimal Product Mix: A set of skincare products for each skin type (oily, combination, normal/dry) across three budget categories (£30, £50, £100).
- Optimal Revenue: The model will output the maximum possible revenue based on the selected product mix.
- The challenge is to curate an optimal selection of skincare products for LOOKFANTASTIC's Boxing Day sale to maximize revenue, while considering customer preferences for skin type (oily, combination, normal/dry) and budget constraints (under £30, under £50, under £100).
- The skincare product data was sourced from a Kaggle dataset and cleaned to include only facial skincare products.
- The expected customer numbers were estimated based on SimilarWeb data for website traffic and Barclays Bank reports on Boxing Day spending.
- Zero-One Integer Programming was used to formulate the problem. The decision variables represent whether a product is included in the optimal mix (1) or not (0).
- Objective function: Maximize total revenue from the selected product mix.
- Constraints: No product repetition across price categories, brand diversity within sets, and skin type distribution across budget ranges.
- The optimization produced nine distinct product sets (one for each combination of skin type and budget range).
- The model achieved a globally optimal solution with a total projected revenue of £319,320.
- Product diversification: Expand the product selection to cater to more specific skin concerns like sensitive or acne-prone skin.
- Seasonal adjustments: Modify product selections based on seasonal skincare needs, such as hydration for winter or sun protection for summer.
- Incorporating customer feedback: Analyze customer reviews and feedback to further refine the product selection.
This project successfully demonstrates the application of Zero-One Integer Programming to optimize product selection for a high-revenue event like Boxing Day. The model balances the need for brand diversity, skin type preferences, and customer budget constraints to maximize revenue.
This project is for academic purposes only as part of the Data Analytics for Strategic Decision Making module.