siokhan / cocktail-maker Goto Github PK

Need to make a classifier, test different types according to estimator flow chart, find which is best.

Implement base markov model, based on example youtube video seen. No user feedback/input yet.

Once estimator has been made, make a script that will generate random cocktails, but then use the estimator to sort between good/bad cocktails.

Vectorize cocktails by converting ingredients to equivalent ingredient ID from ingredientKey csv file.

When cocktail is displayed to, if the user does not like it, reduce its rating by a factor that i need to determine, then, retrain classifier, go again. Need to make sure it doesnt affect good suggestions overall though. Every time a new cocktail is proposed, add it to the db. Also, need to make sure cocktails being produced only use ingredients from googleScraping before i got blocked.

Switch everything to JSON, stream lining everything

Implement proper markov model involving maths and vectorized premade cocktail. Figure out a way to get user input. Also I should set-up a meeting with Thomas.