This project focuses on analyzing a dataset of food and beverage orders to identify coffee-related items and understand common pairings with coffee. The goals are:

1. Determine the sales of coffee-items
2. Developing an upselling feature that recommends items to pair with coffee orders.

• Identification of coffee items using keyword-based filtering and NLP modeling.
• Integration of text and numerical features for enhanced classification.
• Analysis of item pairings with coffee to inform upselling strategies.

The project involves several key steps:

1. Keyword-Based Filtering: Developing a heuristic to identify coffee items based on keywords.
2. Feature Engineering: Combining text data with numerical features like price and quantity for improved model performance.
3. NLP Modeling: Using machine learning models to classify items as coffee or non-coffee based on item names and descriptions.
4. Pairing Analysis: Analyzing order data to determine common pairings with coffee.

The dataset includes various fields such as ORDER_ID, ITEM_NAME, ITEM_DESCRIPTION, ITEM_PRICE, and ITEM_QUANTITY. It represents orders from a food ordering platform, focusing on a single metropolitan area over one month.

• Python 3.x
• Pandas
• Scikit-learn
• Numpy

Jarred Bultema, PhD