parsnip is a R package that offers a unified interface to many machine learning models. By writing an interface between condvis2 and parsnip a vast number of machine learning fits may be explored with condvis.

A list of models supported by parsnip is found on this link: https://www.tidymodels.org/find/parsnip/

Regression

Fit the regression model with parsnip.

library(parsnip)
library(MASS)
library(condvis2)
Boston1 <- Boston[,9:14]

fitlm <-
  linear_reg() %>%
  set_engine("lm") %>%
  fit(medv ~ ., data = Boston1)

fitrf <- rand_forest(mode="regression") %>%
  set_engine("randomForest") %>%
  fit(medv ~ ., data = Boston1)
  

Use condvis to explore the models:

condvis(Boston1, model=list(lm=fitlm,rf=fitrf), response="medv", sectionvars="lstat")

Choose tour “Diff fits” to explore differences between the fits

Some tasks, for example linear regression, support confidence intervals. Tell condvis to plot an interval using pinterval="confidence for that fit. The forest fit does not support confidence intervals so the predictArgs for that fit are NULL.

condvis(Boston1, model=list(lm=fitlm,rf=fitrf), response="medv", sectionvars="lstat",
        predictArgs=list(list(pinterval="confidence"), NULL))

Classification

Fit some classification models:

clmodel <-
   svm_poly(mode="classification") %>%
  set_engine("kernlab") %>%
  fit(Species ~ ., data = iris )

Explore with condvis:

condvis(iris, model=clmodel, response="Species", sectionvars=c("Petal.Length", "Petal.Width"), pointColor="Species")

Click on “Show probs” to see class probabilities.

Survival

Fit a survival model and explore with condvis:

library(survival) # for the data
smodel <-
  surv_reg() %>%
  set_engine("survival") %>%
  fit(Surv(time, status) ~ inst+age+sex+ph.ecog, data=lung)

condvis(na.omit(lung), smodel, response="time", sectionvars = c("inst","sex"), conditionvars=c("age","ph.ecog"))

Clustering

Unlike mlr, parsnip does not yet offer support for clustering fits.