While brainstorming ideas after the theme was announced, we all decided that we wanted to do something impactful, that would affect everyone, and that could change the world for the better. In the end, we decided on saving not only money but also the environment.
Global warming is an increasingly threatening issue, and we need to take steps to prevent our Earth from suffering more. However, we require everyone to play a part to make an impact, and a handful of people still turn a blind eye towards the dangers of climate change.
Therefore, we decided to help the environment by connecting it with something all of us are familiar with and care about - money.
What it does
Energify uses artificial intelligence to predict how much users will spend on environmentally damaging utilities. After users gain insight into how much they spend on environmentally damaging costs, they will realize how much they could save (both money and the environment).
Naturally, users will want to cut down on their monetary-based expenses. As a result, Energify will not only allow its users to cut their costs and spend money on their interests and hobbies, but also effectively combat climate change and save the environment.
How we built it
We trained 5 different TensorFlow neural networks on publicly available data on different household’s monthly average kWh, which we later used to create the forecasting feature for our app based on the user’s household type and their expenses. Then, we made an API with Flask to get the user’s average kWh (kiloWatt hour - a unit commonly used as a billing unit for energy delivered to consumers by electric utilities). Using this, the AI would predict the user's average kWh over the next 5 months. A chart is created based on those values.
The neural network’s predictions are stored using CockroachDB. The predictions are displayed with a beautiful and simple UI made with Svelte and a chatbot in twilio for contact management and appointments directly with the installers.
Challenges we ran into
One of the main challenges that we ran into was collecting the data for our neural networks. Since a household’s average kWh is pretty hard to find, we had to search through different websites until we found a dataset with the historical monthly data of different households. Also, we had to connect a Svelte app to a Flask API, which was also challenging because we had never done that before.
Accomplishments that we're proud of
Training the neural networks and setting up the API for the UI to get the data was probably what we are the most proud of. We are also proud of setting up a Twilio chatbot to contact and set up appointments with us about our app.
What we learned
We learned how to integrate a Flask backend with a Svelte frontend. We also gained more knowledge on training and deploying TensorFlow neural networks with Flask.
What's next for Energify
We plan to help users create a personalized plan and/or tips that will allow users to save more, based on their circumstances. We also plan to get more data for different types of households or even different types of buildings.