Cuisine_Classification_ML

Contains code and artefacts used to attempt identification of the cuisine type of a recipe based on its ingredients, as part of the coursework for the Machine Learning and Predictive Analytics Module for UWE MSc in Data Science.

Dataset

Kaggle - Whats cooking?

Repository structure

• Pre-processing and EDA - Jupyter notebook
• Training and testing - Jupyter notebook
• EDA with Pandas Profiling (hosted on Pages) - https://nutellaweera.github.io/Cuisine_Classification_ML/
• Exploratory scripts (experimentation with parameters, tuning using grid search/trial-and-error) - /scripts folder.
  • Helper/utility methods
  • KNN
  • Naive Bayes
  • Multiclass Logistic Regression
  • Random Forest
  • K-means Clustering
• Graphs and visualizations in the /graphs_and_vis folder

Visualizations and results

Model parameters and performance (avg accuracy scores):

EDA and exploration

Ingredient counts:

Word cloud:

Model training and testing

KNN scores:

Ingredient clusters (k-means):

Confusion matrix: