prediction: Functions for prediction

In ldanai/tidymodel: Tools to build and evalulate preditive models easier

Description

List of prediction functions

Usage

get_confusion_matrix(dta, truth, predicted)

get_important_variables(model, plot = FALSE)

plot_distribution(dta, truth = NULL, ...)

get_prediction(dta, model, predicted, score, target_name)

Arguments

dta	A dataframe that represents model prediction results, score, truth, label to be predicted
truth	A column name (in dta) that represents true labels of the predicted target
predicted	A column name (in dta) that represents predicted classes
model	A ready-built model object
...	A list of selected variables
target_name	A string that represents label name of the predicted target
rank_percentile	A list of percentile to display

Details

• get_prediction Calculates predicted outcomes and propensity scores from the trained model, and Add the prediction columns to original data frame

• plot_distribution generates plots of distributions with classes

• get_confusion_matrix calculates confusion matrix from prediction results. This confusion matrix contains model accuracy, target coverage, lift up value at each rank percentile

• get_important_variables provides the list of important variables from the ready-built model object

Value

• get_confusion_matrix returns a numeric dataframe of confusion matrix.

• get_important_variables returns a list of important variables

• plot_distribution returns a plot of distributions

• get_prediction returns a dataframe with additional columns: predicted outcomes and propensity score

Examples

library(tidymodel)
library(dplyr)

# Show data
as_tibble(dta)

get_important_variables(model)

get_confusion_matrix(dta, TARGET1, PREDICTED)
                            
out = plot_distribution(dta,
                        "TARGET1",
                        "VAR1",
                        "VAR2",
                        "VAR3",
                        "VAR4")
                  
out = plot_distribution(dta,
                        NULL,
                        "VAR1",
                        "VAR2",
                        "VAR3",
                        "VAR4")
                  
# Train data
dta2 = select(dta, TARGET1, VAR1:VAR20)

                      

ldanai/tidymodel documentation built on Jan. 4, 2020, 6:25 a.m.