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:
- Determine the sales of coffee-items
- 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:
- Keyword-Based Filtering: Developing a heuristic to identify coffee items based on keywords.
- Feature Engineering: Combining text data with numerical features like price and quantity for improved model performance.
- NLP Modeling: Using machine learning models to classify items as coffee or non-coffee based on item names and descriptions.
- 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