The lollipop plots the model with the most important interactions and variables in the roots.

# S3 method for lollipop
plot(x, ..., labels = "topAll", log_scale = TRUE,
  threshold = 0.1)

Arguments

x

a result from the lollipop function.

...

other parameters.

labels

if "topAll" then labels for the most important interactions (vertical label) and variables in the roots (horizontal label) will be displayed, if "interactions" then labels for all interactions, if "roots" then labels for all variables in the root.

log_scale

TRUE/FALSE logarithmic scale on the plot. Default TRUE.

threshold

on the plot will occur only labels with Gain higher than `threshold` of the max Gain value in the model. The lower threshold, the more labels on the plot. Range from 0 to 1. Default 0.1.

Value

a ggplot object

Examples

library("EIX") library("Matrix") sm <- sparse.model.matrix(left ~ . - 1, data = HR_data) library("xgboost") param <- list(objective = "binary:logistic", max_depth = 2) xgb_model <- xgboost(sm, params = param, label = HR_data[, left] == 1, nrounds = 25, verbose = 0) lolli <- lollipop(xgb_model, sm) plot(lolli, labels = "topAll", log_scale = TRUE)
library(lightgbm) train_data <- lgb.Dataset(sm, label = HR_data[, left] == 1) params <- list(objective = "binary", max_depth = 3) lgb_model <- lgb.train(params, train_data, 25) lolli <- lollipop(lgb_model, sm) plot(lolli, labels = "topAll", log_scale = TRUE)