plotLocalEffects
Syntax
Description
plotLocalEffects(
creates a bar graph showing the local effects of the terms in the generalized additive model
Mdl,queryPoint)Mdl on the prediction at the specified query point
queryPoint.
plotLocalEffects(
specifies additional options using one or more name-value arguments. For example,
Mdl,queryPoint,Name,Value)'IncludeIntercept',true specifies to include an intercept term in the
bar graph.
b = plotLocalEffects(___)b. Use b to query or modify
Bar Properties of the bar graph after you
create it.
Examples
Train a univariate generalized additive classification model, which contains linear terms for predictors. Classify a new observation using a memory-efficient model object. Then, interpret the prediction for a specified data instance by using the plotLocalEffects function.
Load the ionosphere data set. This data set has 34 predictors and 351 binary responses for radar returns, either bad ('b') or good ('g').
load ionosphereTrain a univariate GAM that identifies whether the radar return is bad ('b') or good ('g').
Mdl = fitcgam(X,Y);
Mdl is a ClassificationGAM model object.
Conserve memory by reducing the size of the trained model.
CMdl = compact(Mdl);
Classify the first observation of the training data, and plot the local effects of the terms in Mdl on the prediction.
label = predict(CMdl,X(1,:))
label = 1×1 cell array
{'g'}
plotLocalEffects(CMdl,X(1,:))
The predict function classifies the first observation X(1,:) as 'g'. The plotLocalEffects function creates a horizontal bar graph that shows the local effects of the 10 most important terms on the prediction. Each local effect value shows the contribution of each term to the classification score for 'g', which is the logit of the posterior probability that the classification is 'g' for the observation.
Train a GAM for binary classification with both linear and interaction terms for predictors. Create local effects plot using both linear and interaction terms in the model, and then create a plot using only linear terms in the model. Specify whether to include interaction terms when creating the local effects plot.
Load the ionosphere data set. This data set has 34 predictors and 351 binary responses for radar returns, either bad ('b') or good ('g').
load ionosphereTrain a GAM using the predictors X and class labels Y. A recommended practice is to specify the class names. Specify to include the 10 most important interaction terms.
Mdl = fitcgam(X,Y,'ClassNames',{'b','g'},'Interactions',10);
Mdl is a ClassificationGAM model object.
Create local effects plots for the 10th observation. Use both the linear and interaction terms in Mdl for the first plot, and use only the linear terms in Mdl for the second plot. To exclude interaction terms, specify 'IncludeInteractions',false.
t = tiledlayout(2,1); title(t,'Local Effects Plots for 10th Observation') nexttile plotLocalEffects(Mdl,X(10,:)) title('GAM with linear and interaction terms') nexttile plotLocalEffects(Mdl,X(10,:),'IncludeInteractions',false) title('GAM with only linear terms')
The plots display the 10 most important terms. Both plots include nine common terms and one uncommon term. The first plot includes the interaction term for x1 and x5, whereas the second plot includes the linear term for x14.
Train a univariate GAM for regression, which contains linear terms for predictors. Then, interpret the prediction for a specified data instance by using the plotLocalEffects function.
Load the data set NYCHousing2015.
load NYCHousing2015The data set includes 10 variables with information on the sales of properties in New York City in 2015. This example uses these variables to analyze the sale prices (SALEPRICE).
Preprocess the data set. Remove outliers, convert the datetime array (SALEDATE) to the month numbers, and move the response variable (SALEPRICE) to the last column.
idx = isoutlier(NYCHousing2015.SALEPRICE); NYCHousing2015(idx,:) = []; NYCHousing2015.SALEDATE = month(NYCHousing2015.SALEDATE); NYCHousing2015 = movevars(NYCHousing2015,'SALEPRICE','After','SALEDATE');
Display the first three rows of the table.
head(NYCHousing2015,3)
BOROUGH NEIGHBORHOOD BUILDINGCLASSCATEGORY RESIDENTIALUNITS COMMERCIALUNITS LANDSQUAREFEET GROSSSQUAREFEET YEARBUILT SALEDATE SALEPRICE
_______ ____________ ____________________________ ________________ _______________ ______________ _______________ _________ ________ _________
2 {'BATHGATE'} {'01 ONE FAMILY DWELLINGS'} 1 0 4750 2619 1899 8 0
2 {'BATHGATE'} {'01 ONE FAMILY DWELLINGS'} 1 0 4750 2619 1899 8 0
2 {'BATHGATE'} {'01 ONE FAMILY DWELLINGS'} 1 1 1287 2528 1899 12 0
Train a univariate GAM for the sale prices. Specify the variables for BOROUGH, NEIGHBORHOOD, BUILDINGCLASSCATEGORY, and SALEDATE as categorical predictors.
Mdl = fitrgam(NYCHousing2015,'SALEPRICE','CategoricalPredictors',[1 2 3 9]);
Mdl is a RegressionGAM model object.
Display the estimated intercept (constant) term of Mdl.
Mdl.Intercept
ans = 3.7518e+05
The intercept term value is close to the average of the response variable in a regression GAM if the training data does not include NaN values. Compute average of the response variable.
mean(NYCHousing2015.SALEPRICE)
ans = 3.7518e+05
Predict the sale price for the first observation of the training data, and plot the local effects of the terms in Mdl on the prediction. Specify 'IncludeIntercept',true to include the intercept term in the plot.
yFit = predict(Mdl,NYCHousing2015(1,:))
yFit = 4.4421e+05
plotLocalEffects(Mdl,NYCHousing2015(1,:),'IncludeIntercept',true)
The predict function predicts the sale price for the first observation as 4.4421e5. The plotLocalEffects function creates a horizontal bar graph that shows the local effects of the terms in Mdl on the prediction. Each local effect value shows the contribution of each term to the predicted sale price.
