Predicting solar panel power generation for everyday and emergency use (hurricanes, etc) on a macro and micro level
- Azure Cloud ML Studio with 84% Accuracy: MaxAbsScaler, XGBoostClassifier, used to classify if output is nominal
- Python with Streamlit and pandas for data ingestion and visualization
- SwiftUI and Figma for presentation app and promotional materials
- Swift/SwiftUI for the dashboard that provides governmental officials and organizations with increased insights into solar energy
- Climate change is driving humanity to more appropriately manage resources, including solar power generation
- Climate change increases the likelihood of hurricanes and other severe weather, solar power generation increases resilience against severe weather, therefore, easier management of solar power is beneficial
MaxAbsScaler
- Scales each feature in the training set by its maximum absolute value
- Each feature here its each columns of the X input matrix
- Estimator scales and translates each feature individually such that the maximal absolute value of each feature is set to 1.0
- Nulls are treated as missing values
- Does not shift/center the data and thus does not destroy any sparsity
XGBoostClassifier
- Extreme Gradient Boosting
- Main driving force behind ML algorithms
- Speed and performance unparalleled and consistently outperforms other algorithms
- Core algorithm can run a cluster of GPUs or even across a network of computers
- Predict one feature based on other studying the instance’s other features
- Implementation of gradient boosted decision trees designed for speed and performance
- Supports both classification and regressive modeling problems
- Ensemble learning and supervised Machine Learning
R2
- 0.95
- How well the data fits the regression model
- Statistical measure in a regression model that determinates the proportion of variance in the dependent variable that can be explained by the independent variable
- Goodness-of-fit, reliability
- Coefficient of determination
