You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
TimeSeries - fix confidence parameter type for some detectors.
- The public API exposed confidence parameters as int even though it's internally implemented as double
- There was no workaround since all classes where double is used are internal
- This caused major issues for software requiring high precision predictions
- This change to API should be backwards compatible since int can be passed to parameter of type double
@@ -233,7 +233,7 @@ public sealed class IidChangePointEstimator : TrivialEstimator<IidChangePointDet
233
233
/// <param name="inputColumnName">Name of column to transform. If set to <see langword="null"/>, the value of the <paramref name="outputColumnName"/> will be used as source.</param>
234
234
/// <param name="martingale">The martingale used for scoring.</param>
235
235
/// <param name="eps">The epsilon parameter for the Power martingale.</param>
@@ -212,7 +212,7 @@ public sealed class IidSpikeEstimator : TrivialEstimator<IidSpikeDetector>
212
212
/// <param name="pvalueHistoryLength">The size of the sliding window for computing the p-value.</param>
213
213
/// <param name="inputColumnName">Name of column to transform. If set to <see langword="null"/>, the value of the <paramref name="outputColumnName"/> will be used as source.</param>
214
214
/// <param name="side">The argument that determines whether to detect positive or negative anomalies, or both.</param>
0 commit comments