mtrglm | R Documentation |
mtrglm() generates different performance metrics and graphs associated with a confusion matrix of a binary variable. It also includes graphs constructed from the predicted values. Although the function is not focused on creating a generalized linear model, it includes options for basic elaboration.
mtrglm(
x,
basic_model = FALSE,
y = NULL,
link = "logit",
partition = 0.75,
seed = 200,
type_ks = 2,
type_pr = 2
)
x |
List that has Values and Probabilities of the response variable in training and test set (the columns must be in this specific order). In case of creating a GLM through the stats::glm() function, you must provide a data frame that includes the response variable and the model covariates. |
basic_model |
Indicates whether a GLM should be created. |
y |
In case of preparing a GLM, this parameter corresponds to the column name of the response variable. |
link |
In case of preparing a GLM, this parameter corresponds to the link function of the binomial family ("logit", "probit", or "cloglog", default is "logit"). |
partition |
In case of preparing a GLM, this parameter corresponds to training set size ratio, default is 0.75. |
seed |
Seed for partition random processes, default is 200. |
type_ks |
Value 0, 1, or 3 indicating the type of KS curve graph that you want to create, look at the |
type_pr |
Value 0, 1, or 3 indicating the type of PR curve graph that you want to create, look at the |
Confusion matrix, Recall, Precision, Accuracy y F1 Score. ROC, PR, KS and Gain curves.
metrics()
to calculate the metrics associated with the confusion matrix,
model_roc()
to plot ROC curve,
model_pr()
to plot Precision - Recall curve,
model_ks()
to plot Kolmogorov - Smirnov curve,
model_gain()
to plot Gain curve.
# example code
sleep_data$Gender = as.numeric(ifelse(sleep_data$Gender == "Male", 0 , 1))
mtrglm(y = "Gender", x = sleep_data, basic_model = TRUE, type_ks = 3, type_pr = 3)
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.