Overview / Usage

By predicting the Solar and Wind energy that can be generated on a particular day, we can reduce the amount of electrical energy that is produced from the non renewable resources because the grid operator will know in advance about the energy demand that will be there so he only has to product that much amount of energy.

Methodology / Approach

This project is still under development as it requires a lot of research, but I have built a prototype for it and below are the details for the same:

• The first thing to do was to research about the factors on which the solar and wind energy depends upon and I found out that there were a lot of them, for both of them.
• Then the harder part was to collect the data which included all the factors. But there was no available dataset that had all of those factors and also I could not find any data which included regions of India. I found a dataset of European countries and decided to use that.
• Then I explored the data and did a lot of cleaning.
• Since a lot of the factors were missing in the data, so I looked for those factors online and manually added them to the dataset.
• Then I trained few different predictive models such as Random Forest, XGBoost etc. and gained a test accuracy of 84%

After training the model I deployed it to an online server and exposed it using an API, then with the help of my friend I created a mobile app and connected it to the API. This app first took information about the location, plant capacity, directions, and some other factors then displayed an interactive plot showing the solar and wind energy that can be produced for the next 7 days.

Technologies Used

I used following technologies and frameworks for this project:

• **Language: **Python
• **Data Analysis and Cleaning: **Pandas and Numpy
• **Plotting: **Matplotlib, Bokeh
• **Model Training: **Scikit-Learn