Split Conformal Prediction
A foundational variant of conformal prediction that constructs prediction intervals by partitioning data into a training set and a calibration set. This approach is straightforward to implement and computationally efficient, making it one of the most commonly used forms of conformal prediction in practice.
Timeline (last 90 days)
In the standard split conformal prediction pipeline, the model is first trained on a training set, then nonconformity scores are computed for each sample on an independent calibration set, and the threshold is taken as a specific quantile of these scores.
Split conformal prediction, also known as inductive conformal prediction, uses the ⌈(1-α)(n+1)⌉/n empirical quantile of the nonconformity scores as the threshold in the calibration step.
The calibration step of split conformal prediction requires no gradient computation, has a time complexity of O(n log n), and can be completed within seconds.
Transductive conformal prediction requires re-traversing the full calibration data for each test sample, making its computational cost far higher than that of split conformal prediction.
It is recommended that developers start with split conformal prediction, as it is simple to implement, computationally efficient, and suitable for rapid validation.
All Facts (5)
In the standard split conformal prediction pipeline, the model is first trained on a training set, then nonconformity scores are computed for each sample on an independent calibration set, and the threshold is taken as a specific quantile of these scores.
50%UnverifiedSplit conformal prediction, also known as inductive conformal prediction, uses the ⌈(1-α)(n+1)⌉/n empirical quantile of the nonconformity scores as the threshold in the calibration step.
50%UnverifiedThe calibration step of split conformal prediction requires no gradient computation, has a time complexity of O(n log n), and can be completed within seconds.
50%UnverifiedTransductive conformal prediction requires re-traversing the full calibration data for each test sample, making its computational cost far higher than that of split conformal prediction.
50%UnverifiedIt is recommended that developers start with split conformal prediction, as it is simple to implement, computationally efficient, and suitable for rapid validation.
50%