Recipe Recommendor

Added a popup which notifies you if the limit for the free api is reached.

Still have to make it look persistently for api responses.

The popup remains hidden for now.

Also have to add some good styling to it.

I migrated the whole website from Python and streamlit to NextJS(hosted on vercel). I redesigned the site and used an API which lets you explore the full recipe!

Made multiple core libraries in order to help me out in “refining” the data as currently it has many inconsistencies in it which alter the results.
vocabulary.py - converts the complex dataset into a unique set of ingredients (a vocabulary)
cleaner.py - cleans the vocabulary into “human-appropriate” ingredients
inspector.py - helps in viewing the data and vocabulary

I also added many tests to check inconsistencies in the vocabularizing algorithm.

I have also put up an under construction notice on the frontend to manage the backend as of now.

Made a vocabulary class which turns the complex list of repeating ingredients in the dataset into a sorted list with unique ingredients.

Also ran repeating tests to put stress on the logic.
It passed all the tests.

I have decided to start from the bottom and do everything from scratch. TF-IDF, Vectorization, Cosine similarity, Data cleansing, Web scraping for ingredient prices, I’ll do all that from scratch.

Created the first working prototype.
I have currently decided to deploy it on streamlit because it is the only way to run the Recommendation Engine and UI together.
Still have t clean the dataset.