This is a simple regressor which is trained on telemetry from a bunch of wind turbines from several different locations. It takes environmental factors such as temperature, wind speeds at different altitudes, wind direction at different altitudes, etc as input & predicts the percentage of the turbine output compared to it's full capacity.
The data was pulled from this Kaggle page.
The project contains the following models trained on the dataset:
- Decision tree
- Random forest
- Linear regression
- Gradient boost
- XG Boost
I have chosen multiple metrics to compare all the models, they are as follows:
- Mean absolute error
- Root mean squared error
- R2
- Adjusted R2
They're visually plotted on graphs corresponding to their scales (R squared & Adj R squared are converted into percentages)
Note: Can you guess the colour theme of the graphs?