ComputeLogisticRegressionStandardDeviation.ComputeStandardDeviation 方法
定義
重要
部分資訊涉及發行前產品，在發行之前可能會有大幅修改。 Microsoft 對此處提供的資訊，不做任何明確或隱含的瑕疵擔保。
計算每個非零訓練權重的標準差矩陣，進一步計算標準差、p 值與 z 分數。 由於 MKL 的規模，這些計算不包含在 Microsoft.ML 套件中。 如果你需要這些計算，請加入 Microsoft.ML.Mkl.Components 套件，並在 Microsoft.ML.Mkl.Components 套件中初始ComputeStandardDeviation化實作ComputeLogisticRegressionStandardDeviation。 由於存在正則化，計算訓練線性係數的變異數時會使用近似方法。
public abstract Microsoft.ML.Data.VBuffer<float> ComputeStandardDeviation(double[] hessian, int[] weightIndices, int parametersCount, int currentWeightsCount, Microsoft.ML.Runtime.IChannel ch, float l2Weight);abstract member ComputeStandardDeviation : double[] * int[] * int * int * Microsoft.ML.Runtime.IChannel * single -> Microsoft.ML.Data.VBuffer<single>Public MustOverride Function ComputeStandardDeviation (hessian As Double(), weightIndices As Integer(), parametersCount As Integer, currentWeightsCount As Integer, ch As IChannel, l2Weight As Single) As VBuffer(Of Single)參數
- hessian
- Double[]
- weightIndices
- Int32[]
- parametersCount
- Int32
- currentWeightsCount
- Int32
- ch
- IChannel
- l2Weight
- Single