Here To Learn - Django & Jupyter Notebook

Django application hosted on Heroku that utilizes a machine learning model trained with over 300,000 data points to predict test outcomes for 5,000 mock students based on eating and sleeping habits for each student.

This Jupyter Notebook demonstrates:

Data generation
JSON API conversion to pandas DataFrame
Data munging (dummy variables for nominal categorical sleep data)
Hyperparameter optimization & model selection in RandomSearchCV
Gradient boosted model training
matplotlib and seaborn visualization

This app is one-third of a Service Oriented Architecture (SOA) app called Here To Learn which integrates with a Rails application which presents the data and serves as the interface for teachers and students and a Sinatra application which stores survey information gathered from students.

Application Highlights:

5 person team project
SOA integration into Rails and Sinatra applications all hosted independently on Heroku
Data munging, hyperparameter optimization & model selection
Matplotlib, seaborn, chart.js data visualization

Contributing

Blake Enyart - Django app, data visualization (chart.js, seaborn, matplotlib), machine learning implementation
William Peterson - Provided input and mentoring on the machine learning model development

Location of Apps in Production

Here to Learn - Rails based central app designed for user interface
Django application - Django based machine learning microservice
Surveys - Sinatra based database microservice

Tech Stack

Data Visualization

Local Configuration for Further Development

Use the following code to setup the code locally:

python3 -m venv env
source env/bin/activate
pip install -r requirements.txt
psql
CREATE DATABASE heretolearn_production;
CREATE USER heretolearn WITH PASSWORD 'badgers';
GRANT ALL PRIVILEGES ON DATABASE heretolearn_production TO heretolearn;
\q
python manage.py migrate
python manage.py runserver

Navigate to localhost:8000 from your browser to see the app in development mode

Django App Demo

Use the following endpoint to see the production app return data from the Sinatra app, make a prediction, and render data and prediction in JSON format:

https://lit-fortress-28598.herokuapp.com/machinelearning/results/?student_id=5