Coffee Recommender

This repository contains notebooks for a personal project on recommending coffee bags using cosine similarity based on numeric attributes and textual descriptions. The project is divided into two main parts: data cleaning and analysis/recommendation. This project uses This Kaggle Dataset by Patrick Klein.

Project Structure

• cleaning.ipynb: Notebook for data cleaning and preprocessing.
• analysis.ipynb: Notebook for implementing the recommendation system and analyzing the results.
• data/: Directory for storing raw and cleaned data files.

Requirements

• Python 3.x
• Jupyter Notebook
• pandas
• scikit-learn

Setup

1. Clone the repository:

   git clone https://github.com/AlanKha/Coffee-Recommender.git
   cd Coffee-Recommender

2. Install the required packages:

   pip install pandas scikit-learn

Usage

1. Run all cells in cleaning.ipynb to generate cleaned_data.csv.

Analysis and Recommendation

Open analysis.ipynb to use the recommendation system. This notebook will:

1. Load the cleaned data.
2. Implement functions to recommend coffee bags based on numeric attributes or textual descriptions.
3. Provide example usage of the recommendation system.

Functions

• recommend_coffees(title, k=5, method='numeric'): Recommends k coffee bags similar to the specified title using either numeric or textual similarity.

Usage

1. Run all cells in analysis.ipynb.

2. Use the recommend_coffees function to get recommendations:

   recommended_coffees_numeric = recommend_coffees('Tropical Summer Colombia La Sierra', k=5, method='numeric')
   print(recommended_coffees_numeric)
   
   recommended_coffees_text = recommend_coffees('Tropical Summer Colombia La Sierra', k=5, method='text')
   print(recommended_coffees_text)

Examples

Data Row Example

	title	rating	acidity_structure	aftertaste	aroma	body	flavor	with_milk	agtron	blind_assessment	bottom_line	coffee_origin	est_price	notes	review_date	roast_level	roaster	roaster_location	url
4	Tropical Summer Colombia La Sierra	93	9	8	9	8	9	nan	60/77	Fruit-driven, crisply chocolaty. Goji berry, dried plum, baking chocolate, amber, narcissus in aroma and cup. Crisply sweet structure with balanced acidity; lightly satiny mouthfeel. Fruit-toned finish supported by notes of baking chocolate.	An experimentally processed Colombia, sweetly fruit-forward with ballast from crisp chocolate notes.	La Sierra, Cauca Department, Colombia	$18.99/8 ounces	Produced by smallholding farmers from trees of the Castillo, Caturra, Pajarito, Tabi and Bourbon varieties of Arabica, and processed by the traditional washed method using species of lactic acid-producing yeast and bacteria during the fermentation step. Merge is a specialty coffee roaster in Harrisonburg, Virginia dedicated to ethical sourcing of high-quality coffees. Visit www.mergecoffeeco.com for more information.	November 2022	Medium-Light	Merge Coffee Company	Harrisonburg, Virginia	https://www.coffeereview.com/review/tropical-summer-colombia-la-sierra/

Notes

• This project is intended for personal use and exploration.
• Adjustments and improvements can be made as needed.