Energy_Predictor

Inspiration

We noticed that many small businesses that start from a local scenario have tough times managing their electricity bills and we wanted to provide a solution to such struggling business owners.

What it does

It looks at 156 weeks of power consumption and runs a LSTM temporal Recurrent Neural Network to understand the data and learn from it. Then it predicts the next 46 weeks of power consumption

How we built it

We used Jupyter Notebooks which ran on Python3. We also used Keras, Numpy and Scikit learn to provide machine learning inputs to the model.

Challenges we ran into

We had difficulty converting the model into a front end usage which would be easily done given more time.

Accomplishments that we're proud of

We could get a test accuracy of 75 percent and a train accuracy of 92 percent. Along with this we can also predict a large timeframe, given our experience in machine learning.

What we learned

We learnt to work in a team and also how to use the Jupyter Notebooks from the technical standpoint. We learnt LSTM RNNs as well along with everything else.