Practical Confidence and Prediction Intervals
By Tom M. Heskes. 1996
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
- Erik Strumbelj
- Igor Kononenko
- Marko Robnik-Sikonja
- Zoran Bosnic
- Hao Quan
- Tim Pearce
- Dipti Srinivasan
- Alexandra Melike Brintrup
Citations in the corpus
- Missing Data Imputation for Supervised Learning (2016, methodology): “For ANNs, prediction intervals are obtained from the standard deviation of test set errors of ANNs trained with different convergences (Heskes 1997).”
- High-Quality Prediction Intervals for Deep Learning: A Distribution-Free, Ensembled Approach (2018, background): “This is an oversimplification (Heskes, 1996), implicitly assuming that training data fully populates the input space, which is seldom the case.”
- Short-Term Load and Wind Power Forecasting Using Neural Network-Based Prediction Intervals (2014, background): “Implementation difficulties, special assumptions about the data distribution, and massive computational requirements [24]–[26] hinder widespread applications of these methods for decision making.”
- Machine learning approaches for estimation of prediction interval for the model output (2006, unknown): “This approach has been widely used in statistical (Harnett & Murphy, 1980; Wonnacott & Wonnacott, 1996) and machine learning communities ( Heskes, 1997; Nix & Weigend, 1994; Penny & Roberts, 1997) for time series forecasting; uncertainty is estimated in terms of confidence interval or prediction in”
- Quantile regression neural networks: Implementation in R and application to precipitation downscaling (2011, background): “While specific functions are not provided, confidence intervals for regression quantiles obtained via ensemble averaging can be estimated following Heskes (1997) and Carney et al. (1999).”
- Explanation and Reliability of Individual Predictions (2013, methodology): “Generalizing the method from [13], we label it BagMLb .”
- Lower Upper Bound Estimation Method for Construction of Neural Network-Based Prediction Intervals (2011, methodology): “By definition, a PI accounts for more sources of uncertainty (model misspecification and noise variance) and is wider than the corresponding CI [12].”