This repository contains code to parse coffee names based on a certain structure, summarize user rating data from a text file, and recommend new coffees to users. This is a coding challenge for TextNow.
To install the dependencies, execute the following commands pip install -r requirements.txt.
The repo contains four executables. The main, coffee.py is called from the command line, and runs parse.py, summarize.py, or collaborative_filtering.py, depending on the user input.
Use parse.py to parse names based on a certain structure:
$ python coffee.py parse "Organic Fair Trade Sweet and Sour Indian"
Decaf False
Organic True
Fair Trade True
Country India
Adjective Sweet And Sour
Use summarize.py to summarize data in a text file:
$ python coffee.py summarize "coffee_ratings.txt"
There are a total of 200 users
There are a total of 189 coffee names
Decaf
False 89
True 100
Fair Trade
False 110
True 79
Organic
False 89
True 100
Adjective
Aa 14
Black Satin 13
Caturra 16
Cuzcachapa 15
Daysiss Water 1
Honey Burst 15
Longberry 10
Mandheling 9
Paradise Valley 18
Peaberry 17
Reserve 12
Sidamo 14
Supremo 18
Swiss Water 17
Country
Bali 16
Bolivia 15
Brazil 15
Costa Rica 10
Dominican Republic 13
El Salvador 16
Ethiopia 9
Guatemala 11
India 16
Kenya 10
Mali 12
Mexico 4
Panama 14
Peru 15
Sumatra 13
Use collaborative_filtering.py to recommend 3 new items based on user data from a text file:
$ python coffee.py recommend "coffee_ratings.txt"
uid coffee name inferred rating
0 daysiss Water Salvadorean 4
0 Organic Fair Trade Decaf Black Satin Bolivian 3
0 Fair Trade Honey Burst Peruvian 3
1 Organic Fair Trade Decaf Black Satin Balinese 4
1 Organic Fair Trade Decaf Paradise Valley Malian 4
1 Organic Fair Trade Decaf Paradise Valley Kenyan 4
2 daysiss Water Salvadorean 4
2 Organic Decaf Sidamo Panamanian 3
2 Organic Decaf AA Guatemalan 3
3 daysiss Water Salvadorean 4
3 Organic Fair Trade Decaf Black Satin Bolivian 3
3 Fair Trade Decaf Swiss Water Sumatran 3
Do a proper train/test split to test the generalizability of the collaborative filtering code. Add unittests for all methods in the repo. And add Travis. Improve comments.