Inge van den Ende-Leveraging conformal prediction for calibrated probabilistic time series forecasts

Calibration of Probabilistic Time Series Forecasts

Conformal Prediction for Calibrated Probabilistic Forecasts

• Conformal prediction is a method for calibrated probabilistic time series forecasts
• It provides a simple and effective way to add uncertainty to point forecasts
• The calibration set is used to determine the correct width of the prediction interval

Quantum Regression for Calibrated Probabilistic Forecasts

• Quantum regression is a method for calibrated probabilistic time series forecasts
• It is a practical and effective way to add uncertainty to point forecasts
• It adapts to the input space and provides a statistical guarantee

Advantages of Conformal Prediction

• Provides a simple and effective way to add uncertainty to point forecasts
• Does not require distributional assumptions
• Provides a statistical guarantee

Disadvantages of Conformal Prediction

• The width of the prediction interval can be narrow or wide depending on the training data
• The method can be computationally expensive for large datasets

Quantum Regression vs. Conformal Prediction

• Quantum regression is a practical and effective way to add uncertainty to point forecasts
• Conformal prediction provides a statistical guarantee, but is more computationally expensive

Calibration of Probabilistic Time Series Forecasts

• Calibration set is used to determine the correct width of the prediction interval
• The calibration set can be randomly selected or selected based on the data distribution

Applications of Conformal Prediction

• Renewable energy transition
• Energy market forecasting
• Load forecasting

Calibration of Probabilistic Time Series Forecasts

• Calibration is a crucial step in probabilistic time series forecasting
• Calibration provides a way to add uncertainty to point forecasts
• Calibration can be done using conformal prediction or quantum regression