Movatterモバイル変換

Type:	Package
Title:	Partial Dependence Plots
Version:	0.8.2
Description:	A general framework for constructing partial dependence (i.e., marginal effect) plots from various types machine learning models in R.
License:	GPL-2 |GPL-3 [expanded from: GPL (≥ 2)]
URL:	https://github.com/bgreenwell/pdp,http://bgreenwell.github.io/pdp/
BugReports:	https://github.com/bgreenwell/pdp/issues
Depends:	R (≥ 3.6.0)
Suggests:	adabag, AmesHousing, C50, caret, covr, Cubist, doParallel,dplyr, e1071, earth, gbm, gridExtra, ICEbox, ipred, keras,kernlab, magrittr, MASS, Matrix, mda, mlbench, nnet, party,partykit, randomForest, ranger, reticulate, rpart, tinytest,xgboost (≥ 0.6-0), knitr, rmarkdown, vip
Imports:	foreach, ggplot2 (≥ 3.0.0), grDevices, lattice, methods,rlang (≥ 0.3.0), stats, utils
LazyData:	TRUE
RoxygenNote:	7.3.2
Encoding:	UTF-8
VignetteBuilder:	knitr
NeedsCompilation:	yes
Packaged:	2024-10-28 17:22:10 UTC; bgreenwell
Author:	Brandon M. Greenwell [aut, cre]
Maintainer:	Brandon M. Greenwell <greenwell.brandon@gmail.com>
Repository:	CRAN
Date/Publication:	2024-10-28 17:50:02 UTC

pdp: A general framework for constructing partial dependence (i.e., marginaleffect) plots from various types machine learning models in R.

Description

Partial dependence plots (PDPs) help visualize the relationship between asubset of the features (typically 1-3) and the response while accounting forthe average effect of the other predictors in the model. They areparticularly effective with black box models like random forests and supportvector machines.

Details

The development version can be found on GitHub: https://github.com/bgreenwell/pdp.As of right now,pdp exports four functions:

• partial - construct partial dependence functions (i.e., objects of class"partial") from various fitted model objects;

• plotPartial - plot partial dependence functions (i.e., objects of class"partial") usinglattice graphics;

• autoplot - plot partial dependence functions (i.e., objects of class"partial") usingggplot2 graphics;

• topPredictors - extract most "important" predictors from various types of fitted models.

Author(s)

Maintainer: Brandon M. Greenwellgreenwell.brandon@gmail.com (ORCID)

See Also

Useful links:

• https://github.com/bgreenwell/pdp

• http://bgreenwell.github.io/pdp/

• Report bugs athttps://github.com/bgreenwell/pdp/issues

Plotting Partial Dependence Functions

Description

Plots partial dependence functions (i.e., marginal effects) usingggplot2 graphics.

Usage

## S3 method for class 'partial'autoplot(  object,  center = FALSE,  plot.pdp = TRUE,  pdp.color = "red",  pdp.size = 1,  pdp.linetype = 1,  rug = FALSE,  smooth = FALSE,  smooth.method = "auto",  smooth.formula = y ~ x,  smooth.span = 0.75,  smooth.method.args = list(),  contour = FALSE,  contour.color = "white",  train = NULL,  xlab = NULL,  ylab = NULL,  main = NULL,  legend.title = "yhat",  ...)## S3 method for class 'ice'autoplot(  object,  center = FALSE,  plot.pdp = TRUE,  pdp.color = "red",  pdp.size = 1,  pdp.linetype = 1,  rug = FALSE,  train = NULL,  xlab = NULL,  ylab = NULL,  main = NULL,  ...)## S3 method for class 'cice'autoplot(  object,  plot.pdp = TRUE,  pdp.color = "red",  pdp.size = 1,  pdp.linetype = 1,  rug = FALSE,  train = NULL,  xlab = NULL,  ylab = NULL,  main = NULL,  ...)

Arguments

Value

A"ggplot" object.

Examples

## Not run: ## Regression example (requires randomForest package to run)## Load required packageslibrary(ggplot2)  # for autoplot() genericlibrary(gridExtra)  # for `grid.arrange()`library(magrittr)  # for forward pipe operator `%>%`library(randomForest)# Fit a random forest to the Boston housing datadata (boston)  # load the boston housing dataset.seed(101)  # for reproducibilityboston.rf <- randomForest(cmedv ~ ., data = boston)# Partial dependence of cmedv on lstatboston.rf %>%  partial(pred.var = "lstat") %>%  autoplot(rug = TRUE, train = boston) + theme_bw()# Partial dependence of cmedv on lstat and rmboston.rf %>%  partial(pred.var = c("lstat", "rm"), chull = TRUE, progress = TRUE) %>%  autoplot(contour = TRUE, legend.title = "cmedv",           option = "B", direction = -1) + theme_bw()# ICE curves and c-ICE curvesage.ice <- partial(boston.rf, pred.var = "lstat", ice = TRUE)grid.arrange(  autoplot(age.ice, alpha = 0.1),                 # ICE curves  autoplot(age.ice, center = TRUE, alpha = 0.1),  # c-ICE curves  ncol = 2)## End(Not run)

Boston Housing Data

Description

Data on median housing values from 506 census tracts in the suburbs of Bostonfrom the 1970 census. This data frame is a corrected version of the originaldata by Harrison and Rubinfeld (1978) with additional spatial information.The data were taken directly fromBostonHousing2 andunneeded columns (i.e., name of town, census tract, and the uncorrectedmedian home value) were removed.

Usage

data(boston)

Format

A data frame with 506 rows and 16 variables.

• lon Longitude of census tract.

• lat Latitude of census tract.

• cmedv Corrected median value of owner-occupied homes in USD 1000's

• crim Per capita crime rate by town.

• zn Proportion of residential land zoned for lots over 25,000 sq.ft.

• indus Proportion of non-retail business acres per town.

• chas Charles River dummy variable (= 1 if tract bounds river; 0 otherwise).

• nox Nitric oxides concentration (parts per 10 million).

• rm Average number of rooms per dwelling.

• age Proportion of owner-occupied units built prior to 1940.

• dis Weighted distances to five Boston employment centers.

• rad Index of accessibility to radial highways.

• tax Full-value property-tax rate per USD 10,000.

• ptratio Pupil-teacher ratio by town.

• b $1000(B - 0.63)^2$ where B is the proportion of blacks by town.

• lstat Percentage of lower status of the population.

References

Harrison, D. and Rubinfeld, D.L. (1978). Hedonic prices and the demand forclean air. Journal of Environmental Economics and Management, 5, 81-102.

Gilley, O.W., and R. Kelley Pace (1996). On the Harrison and Rubinfeld Data.Journal of Environmental Economics and Management, 31, 403-405.

Newman, D.J. & Hettich, S. & Blake, C.L. & Merz, C.J. (1998). UCI Repositoryof machine learning databases[http://www.ics.uci.edu/~mlearn/MLRepository.html] Irvine, CA: University ofCalifornia, Department of Information and Computer Science.

Pace, R. Kelley, and O.W. Gilley (1997). Using the Spatial Configuration ofthe Data to Improve Estimation. Journal of the Real Estate Finance andEconomics, 14, 333-340.

Friedrich Leisch & Evgenia Dimitriadou (2010). mlbench: Machine LearningBenchmark Problems. R package version 2.1-1.

Examples

head(boston)

Exemplar observation

Description

Construct a single "exemplar" record from a data frame. For now, all numericcolumns (including"Date" objects) are replaced with theircorresponding median value and non-numeric columns are replaced with theirmost frequent value.

Usage

exemplar(object)## S3 method for class 'data.frame'exemplar(object)## S3 method for class 'matrix'exemplar(object)## S3 method for class 'dgCMatrix'exemplar(object)

Arguments

Value

A data frame with the same number of columns asobject and asingle row.

Examples

set.seed(1554)  # for reproducibilitytrain <- data.frame(  x = rnorm(100),  y = sample(letters[1L:3L], size = 100, replace = TRUE,             prob = c(0.1, 0.1, 0.8)))exemplar(train)

Partial Dependence Functions

Description

Compute partial dependence functions (i.e., marginal effects) for variousmodel fitting objects.

Usage

partial(object, ...)## Default S3 method:partial(  object,  pred.var,  pred.grid,  pred.fun = NULL,  grid.resolution = NULL,  ice = FALSE,  center = FALSE,  approx = FALSE,  quantiles = FALSE,  probs = 1:9/10,  trim.outliers = FALSE,  type = c("auto", "regression", "classification"),  inv.link = NULL,  which.class = 1L,  prob = FALSE,  recursive = TRUE,  plot = FALSE,  plot.engine = c("lattice", "ggplot2"),  smooth = FALSE,  rug = FALSE,  chull = FALSE,  levelplot = TRUE,  contour = FALSE,  contour.color = "white",  alpha = 1,  train,  cats = NULL,  check.class = TRUE,  progress = FALSE,  parallel = FALSE,  paropts = NULL,  ...)## S3 method for class 'model_fit'partial(object, ...)

Arguments

Value

By default,partial returns an object of classc("data.frame", "partial"). Ifice = TRUE andcenter = FALSE then an object of classc("data.frame", "ice")is returned. Ifice = TRUE andcenter = TRUE then an object ofclassc("data.frame", "cice") is returned. These three classesdetermine the behavior of theplotPartial function which isautomatically called wheneverplot = TRUE. Specifically, whenplot = TRUE, a"trellis" object is returned (seelattice for details); the"trellis" object willalso include an additional attribute,"partial.data", containing thedata displayed in the plot.

Note

In some cases it is difficult forpartial to extract the originaltraining data fromobject. In these cases an error message isdisplayed requesting the user to supply the training data via thetrain argument in the call topartial. In most cases wherepartial can extract the required training data fromobject,it is taken from the same environment in whichpartial is called.Therefore, it is important to not change the training data used to constructobject before callingpartial. This problem is completelyavoided when the training data are passed to thetrain argument in thecall topartial.

It is recommended to callpartial withplot = FALSE and storethe results. This allows for more flexible plotting, and the user will nothave to waste time callingpartial again if the default plot is notsufficient.

It is possible to retrieve the last printed"trellis" object, such asthose produced byplotPartial, usingtrellis.last.object().

Ifice = TRUE or the prediction function given topred.funreturns a prediction for each observation innewdata, then the resultwill be a curve for each observation. These are called individual conditionalexpectation (ICE) curves; see Goldstein et al. (2015) andice for details.

References

J. H. Friedman. Greedy function approximation: A gradient boosting machine.Annals of Statistics,29: 1189-1232, 2001.

Goldstein, A., Kapelner, A., Bleich, J., and Pitkin, E., Peeking Inside theBlack Box: Visualizing Statistical Learning With Plots of IndividualConditional Expectation. (2014)Journal of Computational and GraphicalStatistics,24(1): 44-65, 2015.

Examples

## Not run: ## Regression example (requires randomForest package to run)## Fit a random forest to the boston housing datalibrary(randomForest)data (boston)  # load the boston housing dataset.seed(101)  # for reproducibilityboston.rf <- randomForest(cmedv ~ ., data = boston)# Using randomForest's partialPlot functionpartialPlot(boston.rf, pred.data = boston, x.var = "lstat")# Using pdp's partial functionhead(partial(boston.rf, pred.var = "lstat"))  # returns a data framepartial(boston.rf, pred.var = "lstat", plot = TRUE, rug = TRUE)# The partial function allows for multiple predictorspartial(boston.rf, pred.var = c("lstat", "rm"), grid.resolution = 40,        plot = TRUE, chull = TRUE, progress = TRUE)# The plotPartial function offers more flexible plottingpd <- partial(boston.rf, pred.var = c("lstat", "rm"), grid.resolution = 40)plotPartial(pd, levelplot = FALSE, zlab = "cmedv", drape = TRUE,            colorkey = FALSE, screen = list(z = -20, x = -60))# The autplot function can be used to produce graphics based on ggplot2library(ggplot2)autoplot(pd, contour = TRUE, legend.title = "Partial\ndependence")## Individual conditional expectation (ICE) curves## Use partial to obtain ICE/c-ICE curvesrm.ice <- partial(boston.rf, pred.var = "rm", ice = TRUE)plotPartial(rm.ice, rug = TRUE, train = boston, alpha = 0.2)autoplot(rm.ice, center = TRUE, alpha = 0.2, rug = TRUE, train = boston)## Classification example (requires randomForest package to run)## Fit a random forest to the Pima Indians diabetes datadata (pima)  # load the boston housing dataset.seed(102)  # for reproducibilitypima.rf <- randomForest(diabetes ~ ., data = pima, na.action = na.omit)# Partial dependence of positive test result on glucose (default logit scale)partial(pima.rf, pred.var = "glucose", plot = TRUE, chull = TRUE,        progress = TRUE)# Partial dependence of positive test result on glucose (probability scale)partial(pima.rf, pred.var = "glucose", prob = TRUE, plot = TRUE,        chull = TRUE, progress = TRUE)## End(Not run)

Pima Indians Diabetes Data

Description

Diabetes test results collected by the the US National Institute of Diabetesand Digestive and Kidney Diseases from a population of women who were atleast 21 years old, of Pima Indian heritage, and living near Phoenix,Arizona. The data were taken directly fromPimaIndiansDiabetes2.

Usage

data(pima)

Format

A data frame with 768 observations on 9 variables.

• pregnant Number of times pregnant.

• glucose Plasma glucose concentration (glucose tolerance test).

• pressure Diastolic blood pressure (mm Hg).

• triceps Triceps skin fold thickness (mm).

• insulin 2-Hour serum insulin (mu U/ml).

• mass Body mass index (weight in kg/(height in m)^2).

• pedigree Diabetes pedigree function.

• age Age (years).

• diabetes Factor indicating the diabetes test result (neg/pos).

References

Newman, D.J. & Hettich, S. & Blake, C.L. & Merz, C.J. (1998). UCI Repositoryof machine learning databases[http://www.ics.uci.edu/~mlearn/MLRepository.html]. Irvine, CA: University ofCalifornia, Department of Information and Computer Science.

Brian D. Ripley (1996), Pattern Recognition and Neural Networks, CambridgeUniversity Press, Cambridge.

Grace Whaba, Chong Gu, Yuedong Wang, and Richard Chappell (1995), SoftClassification a.k.a. Risk Estimation via Penalized Log Likelihood andSmoothing Spline Analysis of Variance, in D. H. Wolpert (1995), TheMathematics of Generalization, 331-359, Addison-Wesley, Reading, MA.

Friedrich Leisch & Evgenia Dimitriadou (2010). mlbench: Machine LearningBenchmark Problems. R package version 2.1-1.

Examples

head(pima)

Plotting Partial Dependence Functions

Description

Plots partial dependence functions (i.e., marginal effects) usinglattice graphics.

Usage

plotPartial(object, ...)## S3 method for class 'ice'plotPartial(  object,  center = FALSE,  plot.pdp = TRUE,  pdp.col = "red2",  pdp.lwd = 2,  pdp.lty = 1,  rug = FALSE,  train = NULL,  ...)## S3 method for class 'cice'plotPartial(  object,  plot.pdp = TRUE,  pdp.col = "red2",  pdp.lwd = 2,  pdp.lty = 1,  rug = FALSE,  train = NULL,  ...)## S3 method for class 'partial'plotPartial(  object,  center = FALSE,  plot.pdp = TRUE,  pdp.col = "red2",  pdp.lwd = 2,  pdp.lty = 1,  smooth = FALSE,  rug = FALSE,  chull = FALSE,  levelplot = TRUE,  contour = FALSE,  contour.color = "white",  col.regions = NULL,  number = 4,  overlap = 0.1,  train = NULL,  ...)

Arguments

Examples

## Not run: ## Regression example (requires randomForest package to run)## Load required packageslibrary(gridExtra)  # for `grid.arrange()`library(magrittr)  # for forward pipe operator `%>%`library(randomForest)# Fit a random forest to the Boston housing datadata (boston)  # load the boston housing dataset.seed(101)  # for reproducibilityboston.rf <- randomForest(cmedv ~ ., data = boston)# Partial dependence of cmedv on lstatboston.rf %>%  partial(pred.var = "lstat") %>%  plotPartial(rug = TRUE, train = boston)# Partial dependence of cmedv on lstat and rmboston.rf %>%  partial(pred.var = c("lstat", "rm"), chull = TRUE, progress = TRUE) %>%  plotPartial(contour = TRUE, legend.title = "rm")# ICE curves and c-ICE curvesage.ice <- partial(boston.rf, pred.var = "lstat", ice = TRUE)p1 <- plotPartial(age.ice, alpha = 0.1)p2 <- plotPartial(age.ice, center = TRUE, alpha = 0.1)grid.arrange(p1, p2, ncol = 2)## End(Not run)

Extract Most "Important" Predictors (Experimental)

Description

Extract the most "important" predictors for regression and classificationmodels.

Usage

topPredictors(object, n = 1L, ...)## Default S3 method:topPredictors(object, n = 1L, ...)## S3 method for class 'train'topPredictors(object, n = 1L, ...)

Arguments

Details

This function uses the generic functionvarImp tocalculate variable importance scores for each predictor. After that, they aresorted at the names of then highest scoring predictors are returned.

Examples

## Not run: ## Regression example (requires randomForest package to run)#Load required packageslibrary(ggplot2)library(randomForest)# Fit a random forest to the mtcars datasetdata(mtcars, package = "datasets")set.seed(101)mtcars.rf <- randomForest(mpg ~ ., data = mtcars, mtry = 5, importance = TRUE)# Topfour predictorstop4 <- topPredictors(mtcars.rf, n = 4)# Construct partial dependence functions for top four predictorspd <- NULLfor (i in top4) {  tmp <- partial(mtcars.rf, pred.var = i)  names(tmp) <- c("x", "y")  pd <- rbind(pd,  cbind(tmp, predictor = i))}# Display partial dependence functionsggplot(pd, aes(x, y)) +  geom_line() +  facet_wrap(~ predictor, scales = "free") +  theme_bw() +  ylab("mpg")## End(Not run)

Retrieve the last trellis object

Description

Seetrellis.last.object for more details.

Usage

trellis.last.object(..., prefix)