Analyzing the Epicurious recipe data available on Kaggle using network science and graph theory.
- nltk 3.4.5
- networkx 2.3
- numpy 1.16.5
- pandas 0.25.1
- seaborn 0.9.0
Data Processing and Graph Construction.ipynb - uses natural language processing techniques to extract ingredients from a raw recipe text for each dish and then builds a weighted graph with nodes being dishes and connects them based on the common ingredients they share using Jaccard simialrity index
Exploring the Graph.ipynb - explores the structure and various properties of the graph
Data Exploitation.ipynb - this is the main notebook which tries to answer the research questions. It includes the following:
- Network visualizations
- Clustering using Louvain method (community detection)
- KNN-based recommender system
- Logistic regression model predicting recipe ratings using graph filtering