nuget
This class is a container for arrays and values that are computed during computation of a lasso fit. It also contains the final weights of features.
@author Yasser Ganjisaffar (http://www.ics.uci.edu/~yganjisa/)
LassoFit
01 Syntax
Microsoft.VisualBasic.Data.Bootstrapping.LASSO.LassoFit
02 Methods
| Name | Overloads | Summary |
|---|---|---|
| toNumericTable | 1 | 将 LASSO 的正则化路径导出为统一的二维表对象 NumericTable： 1. |
| getWeights | 1 | |
| ToString | 1 | |
| toDataFrame | 1 |
03 Properties
| Name | Overloads | Summary |
|---|---|---|
| numberOfLambdas | 1 | Number of lambda values |
| intercepts | 1 | Intercepts |
| compressedWeights | 1 | Compressed weights for each solution |
| indices | 1 | Pointers to compressed weights |
| numberOfWeights | 1 | Number of weights for each solution |
| nonZeroWeights | 1 | Number of non-zero weights for each solution |
| lambdas | 1 | The value of lambdas for each solution |
| rsquared | 1 | R^2 value for each solution |
| numberOfPasses | 1 | Total number of passes over data |
| numFeatures | 1 | number of the data features that input from the training data |
| featureNames | 1 |
04 Members
toNumericTable(
String())将 LASSO 的正则化路径导出为统一的二维表对象 NumericTable：
- 每一行对应 lambda 路径上面的一个步骤
- 表的第一列是模型的截距，其余的特征列是各个变量的回归系数
- 标签矩阵之中附带
rsquared、lambdas、Df与numberOfWeights这几个路径指标
Parameters
| Name | Type | Description |
|---|---|---|
featureNames | String() | 特征列的名称。为空的时候使用 LassoFit.featureNames 属性； 数量不匹配的时候自动生成 |
numberOfLambdas
Number of lambda values
intercepts
Intercepts
compressedWeights
Compressed weights for each solution
indices
Pointers to compressed weights
numberOfWeights
Number of weights for each solution
nonZeroWeights
Number of non-zero weights for each solution
lambdas
The value of lambdas for each solution
rsquared
R^2 value for each solution
numberOfPasses
Total number of passes over data
numFeatures
number of the data features that input from the training data
featureNames
getWeights(
Int32)ToString()
toDataFrame()