predictorImportance

Estimates of predictor importance for regression ensemble

expand all in page

Syntax

imp =预测象征（ENS） [imp,ma] = predictorImportance(ens)

描述

小鬼= predictorimportance（ens)计算预测因子重要性的估计值ensby summing these estimates over all weak learners in the ensemble.小鬼has one element for each input predictor in the data used to train this ensemble. A high value indicates that this predictor is important forens。

[小鬼,嘛] =预测象征（ens)返回一个P-经过-P矩阵具有相关性的预测度量P预测指标。

Input Arguments

ens

一种regression ensemble, created byfitrensemble, or by the袖珍的方法。

Output Arguments

`小鬼`	具有与预测变量数（列）相同的元素数量的行矢量`ens。X`。The entries are the estimates of predictor importance, with`0`表示最小的重要性。
`嘛`	一种`P`-经过-`P`预测性衡量标准的矩阵`P`预测指标。元素`马（i，j）`is the predictive measure of association averaged over surrogate splits on predictor`J`对于哪个预测指标`I`是最佳的拆分预测变量。`predictorImportance`平均合奏中所有树木的关联的预测度量。

例子

expand all

Estimate Predictor Importance

打开实时脚本

Estimate the predictor importance for all predictor variables in the data.

加载汽车舞data set.

load汽车舞

种植100棵回归树的合奏MPGusing加速度,Cylinders,Displacement,马力,model_year，和Weight作为预测指标。Specify tree stumps as the weak learners.

x = [加速圆柱体位移马力型模型_year重量];t = Templatetree（'maxnumsplits'，1）;ens = fitrensemble（x，mpg，'Method','LSBoost',“学习者”,t);

估计所有预测变量的预测重要性。

imp =预测象征（ENS）

小鬼=1×60.0150 0 0.0066 0.1111 0.0437 0.5181

Weight, the last predictor, has the most impact on mileage. The second predictor has importance 0, which means that the number of cylinders has no impact on predictions made withens。

预测指标的重要性和Surrogate Splits

打开实时脚本

估计数据中所有变量的预测重要性以及回归树合奏包含替代拆分的位置。

加载汽车舞data set.

load汽车舞

种植100棵回归树的合奏MPGusing加速度,Cylinders,Displacement,马力,model_year，和Weight作为预测指标。将树树桩指定为弱学习者，并确定替代分裂。

x = [加速圆柱体位移马力型模型_year重量];t = Templatetree（'maxnumsplits'，1，'Surrogate','on'）；ens = fitrensemble（x，mpg，'Method','LSBoost',“学习者”,t);

Estimate the predictor importance and predictive measures of association for all predictor variables.

[imp,ma] = predictorImportance(ens)

小鬼=1×60.2141 0.3798 0.4369 0.6498 0.3728 0.5700

嘛=6×61.0000 0.0098 0.0102 0.0098 0.0033 0.0067 0 1.0000 0 0 0 0 0.0056 0.0084 1.0000 0.0078 0.0022 0.0084 0.3537 0.4769 0.5834 1.0000 0.1612 0.5827 0.0061 0.0070 0.0063 0.0064 1.0000 0.0056 0.0154 0.0296 0.0533 0.0447 0.0070 1.0000

比较小鬼to the results inEstimate Predictor Importance,马力has the greatest impact on mileage, withWeight具有第二大影响。

More About

expand all

预测指标的重要性

predictorImportance估计预测指标对整体中每个树学习者的重要性ens并返回加权平均小鬼computed usingens.TrainedWeight。输出小鬼每个预测指标具有一个元素。

predictorImportancecomputes importance measures of the predictors in a tree by summing changes in the node risk due to splits on every predictor, and then dividing the sum by the total number of branch nodes. The change in the node risk is the difference between the risk for the parent node and the total risk for the two children. For example, if a tree splits a parent node (for example, node 1) into two child nodes (for example, nodes 2 and 3), thenpredictorImportanceincreases the importance of the split predictor by

(R₁–R₂–R₃)/N_branch,

whereR_i是节点的节点风险i，和N_branch是分支节点的总数。一种node riskis defined as a node error weighted by the node probability:

R_i=P_iE_i,

whereP_iis the node probability of nodei，和E_i是节点的平均平方误差i。

The estimates of predictor importance depend on whether you use surrogate splits for training.

If you use surrogate splits,predictorImportancesums the changes in the node risk over all splits at each branch node, including surrogate splits. If you do not use surrogate splits, then the function takes the sum over the best splits found at each branch node.
如果您不使用替代分配，则预测器重要性的估计值不取决于预测变量的顺序，但要取决于该顺序。

关联的预测度量

Thepredictive measure of associationis a value that indicates the similarity between decision rules that split observations. Among all possible decision splits that are compared to the optimal split (found by growing the tree), the best替代决定分裂产生关联的最大预测度量。第二好的替代分裂具有第二大相关性预测度量。

Supposex_j和x_kare predictor variablesj和k, respectively, andj≠k。在节点t, the predictive measure of association between the optimal splitx_j<u和a surrogate splitx_k<vis

$λ_{j k} = \frac{min (P_{L}, P_{R}) - (1 - P_{L_{j} L_{k}} - P_{R_{j} R_{k}})}{min (P_{L}, P_{R})} 。$

P_Lis the proportion of observations in nodet，这样x_j<u。下标L代表节点的左孩子t。
P_Ris the proportion of observations in nodet，这样x_j≥u。下标Rstands for the right child of nodet。
$P_{L_{j} L_{k}}$ 是节点上观察值的比例t，这样x_j<u和x_k<v。
$P_{R_{j} R_{k}}$ 是节点上观察值的比例t，这样x_j≥u和x_k≥v。
观察值缺失值x_j或者x_k不要有助于比例计算。

λ_JKis a value in (–∞,1]. Ifλ_JK> 0, thenx_k<vis a worthwhile surrogate split forx_j<u。

算法

元素马（i，j）is the predictive measure of association averaged over surrogate splits on predictorj对于哪个预测指标i是最佳的拆分预测变量。This average is computed by summing positive values of the predictive measure of association over optimal splits on predictori并在预测器上替代拆分j并除以预测变量的最佳拆分总数i, including splits for which the predictive measure of association between predictorsi和jis negative.

Extended Capabilities

GPU Arrays
使用并行计算工具箱™在图形处理单元（GPU）上运行加速代码。

此功能完全支持GPU数组。金宝app有关更多信息，请参阅Run MATLAB Functions on a GPU(Parallel Computing Toolbox)。

predictorImportance

Syntax

描述

Input Arguments

Output Arguments

例子

Estimate Predictor Importance

预测指标的重要性和Surrogate Splits

More About

预测指标的重要性

关联的预测度量

算法

Extended Capabilities

GPU Arrays
使用并行计算工具箱™在图形处理单元（GPU）上运行加速代码。

See Also

Topics

predictorImportance

Syntax

描述

Input Arguments

Output Arguments

例子

Estimate Predictor Importance

预测指标的重要性和Surrogate Splits

More About

预测指标的重要性

关联的预测度量

算法

Extended Capabilities

GPU Arrays使用并行计算工具箱™在图形处理单元（GPU）上运行加速代码。

See Also

Topics

GPU Arrays
使用并行计算工具箱™在图形处理单元（GPU）上运行加速代码。