Heat Maps and Recommenders

BMSTU, linguistics faculty practice, semester 6.

Similarity-based recommendation built on heat maps: items are compared pairwise across several attributes, the resulting similarity matrix is rendered as a heat map, and recommendations fall out of it. Four domain variants share one engine, plus graph and optimisation groundwork.

Structure

Path	Contents
graphs_common	Graph and tree visualisation, BFS and DFS, own
	stack and queue
ants	Travelling salesman: ant colony versus brute force
genetic_algos	Genetic algorithm, two variants
variant-01-restaurants	Restaurant recommender
variant-02-courses	Course recommender
variant-03-courses	Course recommender, second variant
variant-04-cosmetics	Cosmetics recommender

Each variant has the same shape: <domain>_recommender.py for the similarity model, gui_classes.py and app.py for the interface, data/ for the attribute tables.

Run

pip install numpy pandas matplotlib
cd variant-01-restaurants && python main.py

cd ants && python main.py          # ant colony vs exhaustive search
cd graphs_common && python graph.py

Stack

Python, NumPy, pandas, matplotlib