FoodMe: User-Guided Recipe Recommendations

Dataset

https://www.kaggle.com/shuyangli94/food-com-recipes-and-user-interactions

Project Idea

With the vast number of recipes and food blogs online, it can be a real challenge to pick out the recipes that you want to try, and get enough variety in your life. To combat this, we look to build a flexible, user-powered recommendation system for recipes.

Dependencies

Usage has been guaranteed as of Python 3.8

matplotlib
nltk
numpy
pandas
plotly
wordcloud
sklearn

Installation and Use

Installation

We recommend installing in a virtual environment to ensure package compatability.

Installation of the source code simply requires

$ git clone https://github.com/michaelbaluja/ECE143_WI22_FinalProject_Group14.git
$ cd ECE143_WI22_FinalProject_Group14
$ pip install .

Use

To run the notebook code, open Jupyter Notebook and simply run the cells provided. There are helper functions provided through scripts/ for cleaning the data and creating recommendations. To get a recommendation, you can run main.py from scripts/Recommender System.

The data used for development can be found under the data folder. When running, ensure that the proper file path is configured for reading in the data.

File Structure

data/
├── ingr_map.pkl
├── interactions_test.csv
├── interactions_train.csv
├── interactions_validation.csv
├── PP_recipes.csv
├── PP_users.csv
├── RAW_interactions.csv
└── RAW_recipes.csv
notebooks/
├── Data_Vis.ipynb
├── Exploratory_Analysis.ipynb
└── Recipe recommendation based on ingredients.ipynb
scripts/
├── Data Visualization/
│   ├── main.py
│   ├── histogram.py
│   ├── pie_chart.py
│   └── word_cloud.py
├── Recommender System/
│   ├── compute_similarity.py
│   ├── getNewIngredients.py
│   ├── load_data.py
│   ├── main.py
│   └── visualization.py
├── PDV Analysis/
│   └── final_pdv_analysis.py
└── High Rating Reason/
    └── why_high_rating.py
presentation/
└── ECE143_Group14.pptx.pdf
README.md
setup.py