DS

Data Science repository for BW-Med-Cab-2. Includes API calls and interaction instructions. Also includes steps to re-create API locally, including dependency installations.

Purpose

The purpose of this API is to recommend medical cannabis strains to patients, using a Nearest Neighbors model trained on the medical effects of more than 1000 cannabis strains.

This API includes a PostgreSQL database connection populated with data from the following sources:

• Kaggle: https://www.kaggle.com/kingburrito666/cannabis-strains
• The Strain: http://strains.evanbusse.com/index.html

Strain Recommender tool receives input of medical ailments/symptoms (ex. stress, anxiety, insomnia, eye pressure) and returns a recommended cannabis strain (ex. King Cake) with it's attributes (name, flavors, rating, etc.).

---

Heroku App: https://med-cab-1415.herokuapp.com/

Endpoints if deployed to Heroku:
Below are the routes that return key-value pair data in JSON from a postgreSQL database.

https://med-cab-1415.herokuapp.com/
https://med-cab-1415.herokuapp.com/strains
https://med-cab-1415.herokuapp.com/recx
https://med-cab-1415.herokuapp.com/model
https://med-cab-1415.herokuapp.com/model/<symptoms_string>
https://med-cab-1415.herokuapp.com/data
https://med-cab-1415.herokuapp.com/toptenrating
https://med-cab-1415.herokuapp.com/toptenflavor

Heroku App II: https://greensolx2.herokuapp.com/

Endpoints if deployed to Heroku:
Below are the routes that return key-value pair data in JSON from a postgreSQL database.

https://greensolx2.herokuapp.com/
https://greensolx2.herokuapp.com/strains
https://greensolx2.herokuapp.com/recx
https://greensolx2.herokuapp.com/model
https://greensolx2.herokuapp.com/model/<symptoms_string>
https://greensolx2.herokuapp.com/data
https://greensolx2.herokuapp.com/toptenrating
https://greensolx2.herokuapp.com/toptenflavor

Note on dual APIs: Both APIs provide same functionality, use same model, and can be used interchangeably. Presence of both is for development and production workflow, as well as back up API.

---

MedCab API

• Strain Recommender Tool
  • To get strain recommendations, type/insert a "symptoms string" into:
    • https://med-cab-1415.herokuapp.com/model/<symptoms_string>
    • or:
    • https://greensolx2.herokuapp.com/model/<symptoms_string>
    • Replace <symptoms_string> with ailments/symptoms
  • Examples:
    • Single Entry:

    https://med-cab-1415.herokuapp.com/model/insomnia
    

    • Two Inputs: (can type in commas and spaces)

    https://med-cab-1415.herokuapp.com/model/insomnia, anxiety
    

    • Multi-Input:

    https://med-cab-1415.herokuapp.com/model/insomnia, anxiety, fatigue, spasms, muscle pain
    

  • Output (json key value pairs)
    • Schema:

    --strain        (strain name, string)
    --id            (strain id, int)
    --flavors       (flavors, string)
    --effects       (positive effects, string)
    --medical       (medical effects, string)
    --type          (indica, hybrid or sativa, string)
    --rating        (up to 5 stars, float)
    

    • Example:

    {"strain":"Crystal Gayle","id":634,"flavors":"Earthy, Diesel, Skunk","effects":"Hungry, Euphoric, Happy, Creative, Focused","medical":"Muscle Spasms","type":"hybrid","rating":4.4}
    

• Routes for data grabs / queries:
  • View all strains in database https://med-cab-1415.herokuapp.com/strains

  # Returns strain id, name, and rating

  • View all data in database https://med-cab-1415.herokuapp.com/data

  # Returns strain name, id, flavors, effects, medical, type, rating, flavor

  • View top ten highest rated strains https://med-cab-1415.herokuapp.com/toptenrating

  # Returns strain names, sorted in descending order, filtered by star rating and length of "medical" description

  • View top ten "most flavorful" strains https://med-cab-1415.herokuapp.com/toptenflavor

  # Returns strain names, sorted in descending order, filtered by length of "flavors" description

---

Resources

• PostgreSQL
• ElephantSQL
• Flask
• Heroku
• scikit-learn

---

Installation Instructions for Dependencies

Flask, Flask-Cors, Psycopg2, Gunicorn, Requests, Dotenv

• pipenv

pipenv install Flask flask-cors psycopg2-binary gunicorn requests python-dotenv

• conda

pip install Flask flask-cors psycopg2-binary gunicorn requests python-dotenv

• PostgreSQL Database Connection Example of format to place credentials inside a .env file:

DB_USER="___________"
DB_NAME="___________"
DB_PASSWORD="___________"
DB_HOST="___________"

---

Running the app locally using Flask

In a terminal:
Mac/Linux:
FLASK_APP=MedCab flask run
Windows:
export FLASK_APP=MedCab (set env var)
flask run

---

Heroku Deployment

• Add "Procfile" (case-sensitive) with following content:

web: gunicorn "MedCab:create_app()"

• Log in to Heroku from the CLI (first time only):

heroku login

• Creating a new application server (MUST BE DONE FROM WITHIN THE REPOSITORY'S ROOT DIRECTORY):

git remote -v
heroku create # optionally provide a name... "heroku create my-app-name"
git remote -v

• Deploying to Production:

git push heroku master

• Viewing production app in browser:

heroku open

• Checking production server logs:

heroku logs --tail

• Configuring production environment variables:

heroku config:set DB_USER="___________"
heroku config:set DB_NAME="___________"
heroku config:set DB_PASSWORD="___________"
heroku config:set DB_HOST="___________"