Skip to content

Prediction and Analysis of Household Energy Consumption - Finalist for IBM Hack Challenge 2019

License

Notifications You must be signed in to change notification settings

adityavinodk/energy_save

Repository files navigation

EnergySave Logo


Problem Statement

The aim of this project is to build Machine Learning models to create energy consumption profiles for household and identify probable areas to plug wastage of energy in household.

The objective of this application is to help households in 2 main ways -

  1. Predict the Energy Star Rating of old or vampire appliances without a energy star rating
  2. Predict reasons for high energy consumption using the valuable insights provided

Starting App

Firstly, add chromedriver to your system path. This is needed to ensure complete advantage of the UI/UX we offer.

Run the following in the root folder of the repository

pip install -r requirements.txt
cd frontend
npm install
npm run build
cd ../server
python app.py

Then open localhost:5000 your browser to start the application. Test the API's using the inputs from sample_inputs.txt file.

About the Project

We have used Jupyter Notebook for visualization and making inferences of the data. We are running the web application using Flask server and using React for the front-end of the application. We have REST API calls to all the predictions are made currently using simple supervised learning classifier models and average around 85% accuracy.

  • datasets folder consists of all the datasets along with their label files
  • data_infereces folder consists of Jupyter files containing the prediction as well as the different inferences made based on the data from the datasets
  • frontend folder consists the source files of the front-end of the web application
  • server consists of all the server code, along with the REST API for making the predictions based on the input data

Team Members

OpenSource Contribution

We would like the developer community to contribute to this project -

  • Move to real-time data prediction, which will be easier to scale in the future
  • Add more useful datasets which provide a larger idea of the usage of the appliances at households like seasonal data.
  • Provide visualitzation of the inferences made on the website through charts and plots.

Send Pull Requests and we shall review them!