predictorImportance

Estimates of predictor importance for regression ensemble

Syntax

imp = predictorImportance(ens) [imp,ma] = predictorImportance(ens)

Description

imp = predictorImportance(ens) computes estimates of predictor importance for ens by summing these estimates over all weak learners in the ensemble. imp has one element for each input predictor in the data used to train this ensemble. A high value indicates that this predictor is important for ens.

[imp,ma] = predictorImportance(ens) returns a P-by-P matrix with predictive measures of association for P predictors.

Input Arguments

ens

A regression ensemble, created by fitrensemble, or by the compact method.

Output Arguments

`imp`	A row vector with the same number of elements as the number of predictors (columns) in `ens.X`. The entries are the estimates of predictor importance, with `0` representing the smallest possible importance.
`ma`	A `P`-by-`P` matrix of predictive measures of association for `P` predictors. Element `ma(I,J)` is the predictive measure of association averaged over surrogate splits on predictor `J` for which predictor `I` is the optimal split predictor. `predictorImportance` averages this predictive measure of association over all trees in the ensemble.

Examples

expand all

Estimate Predictor Importance

Open Live Script

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

Load the carsmall data set.

load carsmall

Grow an ensemble of 100 regression trees for MPG using Acceleration, Cylinders, Displacement, Horsepower, Model_Year, and Weight as predictors. Specify tree stumps as the weak learners.

X = [Acceleration Cylinders Displacement Horsepower Model_Year Weight];
t = templateTree('MaxNumSplits',1);
ens = fitrensemble(X,MPG,'Method','LSBoost','Learners',t);

Estimate the predictor importance for all predictor variables.

imp = predictorImportance(ens)

imp = 1×6

    0.0150         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 with ens.

Predictor Importance and Surrogate Splits

Open Live Script

Estimate the predictor importance for all variables in the data and where the regression tree ensemble contains surrogate splits.

Load the carsmall data set.

load carsmall

Grow an ensemble of 100 regression trees for MPG using Acceleration, Cylinders, Displacement, Horsepower, Model_Year, and Weight as predictors. Specify tree stumps as the weak learners, and also identify surrogate splits.

X = [Acceleration Cylinders Displacement Horsepower Model_Year Weight];
t = templateTree('MaxNumSplits',1,'Surrogate','on');
ens = fitrensemble(X,MPG,'Method','LSBoost','Learners',t);

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

[imp,ma] = predictorImportance(ens)

imp = 1×6

    0.2143    0.3779    0.4282    0.6585    0.3674    0.5624

ma = 6×6

    1.0000    0.0098    0.0102    0.0098    0.0033    0.0067
         0    1.0000         0         0         0         0
         0         0    1.0000         0         0         0
    0.2638    0.3036    0.3555    1.0000    0.1045    0.3934
    0.0051    0.0050    0.0047    0.0051    1.0000    0.0040
    0.0146    0.0253    0.0370    0.0328    0.0070    1.0000

Comparing imp to the results in Estimate Predictor Importance, Horsepower has the greatest impact on mileage, with Weight having the second greatest impact.

More About

expand all

Predictor Importance

predictorImportance computes estimates of predictor importance for ens by summing changes in the mean squared error (MSE) due to splits on every predictor and dividing the sum by the number of branch nodes. If ens is grown without surrogate splits, this sum is taken over best splits found at each branch node. If ens is grown with surrogate splits, this sum is taken over all splits at each branch node including surrogate splits. imp has one element for each input predictor in the data used to train ens. At each node, MSE is estimated as node error weighted by the node probability. Variable importance associated with this split is computed as the difference between MSE for the parent node and the total MSE for the two children.

Predictive Measure of Association

The predictive measure of association is 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 surrogate decision split yields the maximum predictive measure of association. The second-best surrogate split has the second-largest predictive measure of association.

Suppose x_j and x_k are predictor variables j and k, respectively, and j ≠ k. At node t, the predictive measure of association between the optimal split x_j < u and a surrogate split x_k < v is

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

P_L is the proportion of observations in node t, such that x_j < u. The subscript L stands for the left child of node t.
P_R is the proportion of observations in node t, such that x_j ≥ u. The subscript R stands for the right child of node t.
$P_{L_{j} L_{k}}$ is the proportion of observations at node t, such that x_j < u and x_k < v.
$P_{R_{j} R_{k}}$ is the proportion of observations at node t, such that x_j ≥ u and x_k ≥ v.
Observations with missing values for x_j or x_k do not contribute to the proportion calculations.

λ_jk is a value in (–∞,1]. If λ_jk > 0, then x_k < v is a worthwhile surrogate split for x_j < u.

Algorithms

Element ma(i,j) is the predictive measure of association averaged over surrogate splits on predictor j for which predictor i is the optimal split predictor. This average is computed by summing positive values of the predictive measure of association over optimal splits on predictor i and surrogate splits on predictor j and dividing by the total number of optimal splits on predictor i, including splits for which the predictive measure of association between predictors i and j is negative.

Documentation