modelDown: Create page with summary for your models

In kromash/modelDown: Make Static HTML Website for Predictive Models

Description

Create page with summary for your models

Usage

modelDown(..., modules = c("model_performance", "variable_importance",
  "variable_response", "prediction_breakdown"), output_folder = "output")

Arguments

...	explainers
modules	selected modules to generate
output_folder	folder where generated page will be stored

Details

Additional argument that could by passed by name:

• vr.vars variables which will be examined in Variable Response module. Defaults to all variables. Example vr.vars = c("var1", "var2")

• pb.observations observations which will be examined in Prediction Breakdown module. When not given it selects worst predicted observations for each model. Example pb.observations = c(1,2,3) where 1,2,3 are observation numbers.

• vr.type types of examinations which will be conducteed in Variable Response module. Defaults to c("ale", "pdb"). Example vr.type = "ale"

• plot_width default width for plots. Defaults to 800. Example plot_width = 750

• vr.plot_width Override plot width for Variable Response module. Defaults to plot_width. Example vr.plot_width = 750

• mp.plot_width Override plot width for Model Performance module. Defaults to plot_width. Example mp.plot_width = 750

• pb.plot_width Override plot width for Prediction Breakdown module. Defaults to plot_width. Example pb.plot_width = 750

• vi.plot_width Override plot width for Variable Importance module. Defaults to plot_width. Example vi.plot_width = 750

Author(s)

Magda Tatarynowicz, Kamil Romaszko, Mateusz Urab<c5><84>ski

Examples

## Not run: 
require(ranger)
require(breakDown)
require(DALEX)

# ranger
HR_ranger_model <- ranger(as.factor(left) ~ .,
                      data = HR_data, num.trees = 500, classification = TRUE, probability = TRUE)
explainer_ranger <- explain(HR_ranger_model,
                      data = HR_data, y = HR_data$left, function(model, data) {
 return(predict(model, data)$prediction[,2])
}, na.rm=TRUE)

# glm
HR_glm_model <- glm(left~., HR_data, family = "binomial")
explainer_glm <- explain(HR_glm_model, data=HR_data, y = HR_data$left)

modelDown::modelDown(explainer_ranger, explainer_glm) #all defaults

modelDown::modelDown(explainer_glm,
  modules = c("model_performance", "variable_importance",
              "variable_response", "prediction_breakdown"),
  output_folder = "modelDown_output",
  vr.vars= c("average_montly_hours", "time_spend_company"),
  pb.observations = c(1,2,3),
  vr.type = "ale",
  plot_width=700,
  pb.plot_width=810,
  mp.plot_width=820,
  vi.plot_width=830,
  vr.plot_width=840)

## End(Not run)

kromash/modelDown documentation built on May 23, 2019, 1:13 p.m.