R/assess_models.R
In nlanderson9/languagePredictR: Predict Outcomes from Natural Language
Defines functions
assess_models
Documented in
assess_models
#' modelAssessment Class
#'
#' @slot type Model type, "binary" or "continuous"
#' @slot model_labels Label names and basic info for each model in the assessment
#' @slot auc_polygon_df Dataframe for plotting the AUC polygon
#' @slot roc_ci_df Dataframe for plotting the ROC confidence intervals
#' @slot roc_curve_df Dataframe for plotting the ROC curves
#' @slot auc_ci_labels_df Dataframe for AUC labels
#' @slot auc_tests Dataframe of significance tests for AUCs
#' @slot cat_data Dataframe of predictors (word engrams) with their model weights
#'
#' @export modelAssessment
#' @exportClass modelAssessment
#'
modelAssessment = setClass("modelAssessment", slots = c("type", "model_labels", "auc_polygon_df", "roc_ci_df", "roc_curve_df", "auc_ci_labels_df", "auc_tests", "cat_data"))
#' @title Create Model Assessment
#'
#' @description This function is deprecated; all dependent functions (e.g. \code{plot_roc()} or \code{plot_predictor_words}) now take individual models as arguments. This function assesses one or more models created by the \code{\link{language_model}} function.
#'
#' @param ... Models generated by the \code{\link{language_model}} function, and/or two-column dataframes with a predictor variable and an outcome variable
#'
#' @return An object of the type "modelAssessment"
#'
#' @import pROC
#' @importFrom methods setClass new
#' @importFrom rlang .data
#'
#' @export
#'
#' @examples
#'
#' \dontrun{
#' strong_movie_review_data$cleanText = clean_text(strong_movie_review_data$text)
#' mild_movie_review_data$cleanText = clean_text(mild_movie_review_data$text)
#'
#' # Using language to predict "Positive" vs. "Negative" reviews
#' # Only for strong reviews (ratings of 1 or 10)
#' movie_model_strong = language_model(strong_movie_review_data,
#' outcomeVariableColumnName = "valence",
#' outcomeVariableType = "binary",
#' textColumnName = "cleanText")
#'
#' # Using language to predict "Positive" vs. "Negative" reviews
#' # Only for mild reviews (ratings of 4 or 7)
#' movie_model_mild = language_model(mild_movie_review_data,
#' outcomeVariableColumnName = "valence",
#' outcomeVariableType = "binary",
#' textColumnName = "cleanText")
#'
#' # Create the model assessment
#' # movie_assessment = assess_models(movie_model_strong, movie_model_mild)
#' }
#'
#' @details
#' The primary purpose of this function is to be used with other functions included in this package, such as plot_roc() or predictor_word_plots()
#' All necessary calculations are performed by this function, so output plots and analyses can be performed quickly and modified as needed
#' This function can be used to assess models generated by the \code{\link{language_model}}, as well as simple predictors that could be compared with language models.
assess_models = function(...) {
dots = list(...)
dots_names = match.call(expand.dots = FALSE)
model_type = c()
for (i in 1:length(dots)) {
input = dots[[i]]
if (class(input) == "langModel") {
model_type = c(model_type, input@type)
}
else if (is.data.frame(input)) {
if (ncol(input) != 2) {
stop("Your dataframe '", as.character(dots_names$...[[i]]), "' should have 2 columns: a column of predictor variables, and a column of outcome variables.")
}
column_names = colnames(input)
if (is.factor(input[[column_names[2]]]) & nlevels(input[[column_names[2]]]) == 2) {
model_type = c(model_type, "binary")
}
else if (is.numeric(input[[column_names[2]]])) {
model_type = c(model_type, "continuous")
}
else {
stop(paste("The second (outcome) column of your dataframe '", as.character(dots_names$...[[i]]), "' must be a 2-level factor or numeric variable.", sep=""))
}
}
else {
stop(paste("Your argument '", as.character(dots_names$...[[i]]), "'must be either a langModel generated by the create_lasso_regression() function, or a two-column dataframe.", sep=""))
}
}
if (length(unique(model_type)) != 1) {
stop("All models being assessed must predict the same type of variable (binary outcomes or continuous outcomes).")
}
else {
model_type = model_type[1]
}
if (model_type == "binary") {
results_matrix = data.frame(matrix(nrow=0, ncol=8))
colnames(results_matrix) = c("model", "auc", "auc_ci", "predictive_acc", "high_predictor_1", "high_predictor_2", "low_predictor_1", "low_predictor_2")
model_labels = data.frame(matrix(ncol=7,nrow=0))
colnames(model_labels) = c("name","LASSO_model", "auc", "ci_lower", "ci_upper", "cat0title", "cat1title")
roc_plot_data = data.frame(matrix(ncol=2,nrow=0))
colnames(roc_plot_data) = c("specificities", "sensitivities")
roc_ci_plot_data = data.frame(matrix(ncol=4, nrow=0))
colnames(roc_ci_plot_data) = c("percent2p5", "percent50", "percent97p5", "sensitivities")
roc_list = list()
cat_data = data.frame(matrix(ncol=4,nrow=0))
colnames(cat_data) = c("words", "weights", "class", "model")
for (i in 1:length(dots)) {
input = dots[[i]]
if (class(input) == "langModel") {
LASSO_model = TRUE
predictor<- as.numeric(predict(input@cv,newx=input@x,s=input@lambda))
roc_data <- suppressMessages(roc(input@y,predictor, ci=TRUE))
roc_list[[as.character(dots_names$...[[i]])]] = roc_data
roc_ci_data = ci.sp(roc_data, sensitivities=seq(0,1,.01))
if (input@type == "binary") {
cat1title = paste('Words predicting\n"',input@level1, '" responses', sep="")
cat0title = paste('Words predicting\n"',input@level0, '" responses', sep="")
}
else {
cat1title = paste('Words predicting\n"high-',input@outcome, '" responses', sep="")
cat0title = paste('Words predicting\n"low-',input@outcome, '" responses', sep="")
}
cat0data = input@cat0raw
cat1data = input@cat1raw
if (nrow(cat0data) == 0) {
cat0data = data.frame(words="", weights=0)
}
if (nrow(cat1data) == 0) {
cat1data = data.frame(words="", weights=0)
}
cat0data$class = "cat0"
cat1data$class = "cat1"
cat_data_temp = rbind(cat0data, cat1data)
cat_data_temp$model = as.character(dots_names$...[[i]])
cat_data = rbind(cat_data, cat_data_temp)
}
else {
LASSO_model = FALSE
column_names = colnames(input)
input1_x = input[[column_names[1]]]
response = input[[column_names[2]]]
predictor_title1 = paste("'",column_names[1],"'", sep="")
response_title1 = paste("'",column_names[2],"'", sep="")
glm1<-glm(response~input1_x,family="binomial")
predictor = as.numeric(predict(glm1,newx=input1_x))
roc_data <- suppressMessages(roc(response,predictor, ci=TRUE))
roc_list[[as.character(dots_names$...[[i]])]] = roc_data
roc_ci_data = ci.sp(roc_data, sensitivities=seq(0,1,.01))
cat1title = ""
cat0title = ""
cat_data_temp = data.frame(words="", weights = 0, class=NA, model = as.character(dots_names$...[[i]]))
cat_data = rbind(cat_data, cat_data_temp)
}
roc_data_formatted = data.frame(specificities=roc_data$specificities, sensitivities=roc_data$sensitivities)
roc_data_formatted$model = as.character(dots_names$...[[i]])
roc_ci_data_formatted = as.data.frame(roc_ci_data)
roc_ci_data_formatted$sensitivities = as.numeric(rownames(roc_ci_data_formatted))
colnames(roc_ci_data_formatted) = c("percent2p5", "percent50", "percent97p5", "sensitivities")
roc_ci_data_formatted$model = as.character(dots_names$...[[i]])
roc_plot_data = rbind(roc_plot_data, roc_data_formatted)
roc_ci_plot_data = rbind(roc_ci_plot_data, roc_ci_data_formatted)
temp_frame = data.frame(name=as.character(dots_names$...[[i]]), LASSO_model=LASSO_model, auc = roc_data$auc, ci_lower = roc_data$ci[1], ci_upper = roc_data$ci[3], cat0title = cat0title, cat1title=cat1title)
model_labels = rbind(model_labels, temp_frame)
}
roc_plot_data$model = factor(roc_plot_data$model, levels = model_labels$name)
roc_ci_plot_data$model = factor(roc_ci_plot_data$model, levels = model_labels$name)
lowest_auc_model = model_labels$name[which.min(model_labels$auc)]
box_df = data.frame(polygon.x = c(0,0,1,1,0), polygon.y = c(0,1,1,0,0))
auc_df = data.frame(matrix(ncol=3,nrow=0))
colnames(auc_df) = c("sensitivities", "specificities", "model")
for (j in 1:nrow(model_labels)) {
auc_df_add = subset(roc_plot_data, model == model_labels$name[j])
auc_extra = data.frame(sensitivities = c(1,0,0), specificities = c(0,0,1), model=rep(model_labels$name[j],3))
auc_df = rbind(auc_df, auc_df_add, auc_extra)
}
auc_labels = data.frame(matrix(ncol=4,nrow=0))
colnames(auc_labels) = c("model", "label_text", "x", "y")
for (j in 1:nrow(model_labels)) {
model = model_labels$name[j]
label_text = paste("AUC: ",round(model_labels$auc[j],3), "\nCI: (", round(model_labels$ci_lower[j],3), "-", round(model_labels$ci_upper[j],3),")",sep="")
temp_frame = data.frame(model=model, label_text=label_text, x=.4, y=.6-(.1*j))
auc_labels = rbind(auc_labels, temp_frame)
}
roc_plot_data_all = CreateAllFacet(roc_plot_data, "model")
roc_ci_plot_data_all = CreateAllFacet(roc_ci_plot_data, "model")
auc_df$facet = auc_df$model
auc_df_add = subset(roc_plot_data, model == lowest_auc_model)
auc_df_add$facet = "all"
auc_extra = data.frame(sensitivities = c(1,0,0), specificities = c(0,0,1), model=rep(model_labels$name[j],3), facet=rep("all",3))
auc_df = rbind(auc_df, auc_df_add, auc_extra)
roc_plot_data_all$facet = factor(roc_plot_data_all$facet, levels = c(model_labels$name, "all"))
roc_ci_plot_data_all$facet = factor(roc_ci_plot_data_all$facet, levels = c(model_labels$name, "all"))
auc_df$facet = factor(auc_df$facet, levels = c(model_labels$name, "all"))
auc_labels$facet = "all"
auc_labels_new = auc_labels
auc_labels_new$facet = auc_labels_new$model
auc_labels_new$y=max(auc_labels_new$y)
auc_labels = rbind(auc_labels_new, auc_labels)
auc_labels$facet = factor(auc_labels$facet, levels = c(model_labels$name, "all"))
auc_tests = data.frame(matrix(ncol=5,nrow=0))
colnames(auc_tests) = c("model1", "model2", "model1_auc", "model2_auc", "p_value")
for (i in 1:length(roc_list)) {
for (j in (i+1):length(roc_list)) {
if (j > length(roc_list)) {
next
}
model1 = names(roc_list)[i]
model2 = names(roc_list)[j]
if (model1 == model2) {
next
}
model1_auc = roc_list[[i]]$auc
model2_auc = roc_list[[j]]$auc
p_value = roc.test(roc_list[[i]], roc_list[[j]])
temp_frame = data.frame(model1=model1, model2=model2, model1_auc=model1_auc, model2_auc=model2_auc, p_value=p_value$p.value)
auc_tests = rbind(auc_tests, temp_frame)
}
}
auc_tests$sig_TF = ifelse(auc_tests$p_value < .05, 1, 0)
auc_tests$sig_TF = factor(auc_tests$sig_TF, levels = c(0,1))
auc_tests$font = ifelse(auc_tests$sig_TF == 0, "plain", "bold")
auc_tests$size = ifelse(auc_tests$sig_TF == 0, 0, 1)
auc_tests$size = as.factor(auc_tests$size)
auc_tests$width = ifelse(auc_tests$sig_TF == 0, 1, .95)
auc_tests$height = ifelse(auc_tests$sig_TF == 0, 1, .95)
auc_tests$sig = ifelse(auc_tests$p_value >= .05, "NS", ifelse(auc_tests$p_value >= .01, "*", ifelse(auc_tests$p_value >= .001, "**", "***")))
auc_tests$model1 = factor(auc_tests$model1, levels = model_labels$name)
auc_tests$model2 = factor(auc_tests$model2, levels = rev(model_labels$name))
output = new("modelAssessment", type="binary", model_labels=model_labels, auc_polygon_df=auc_df, roc_ci_df=roc_ci_plot_data_all, roc_curve_df=roc_plot_data_all, auc_ci_labels_df=auc_labels, auc_tests=auc_tests, cat_data=cat_data)
return(output)
}
}
nlanderson9/languagePredictR documentation built on June 10, 2021, 11 a.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Defines functions assess_models
Documented in assess_models
#' modelAssessment Class
#'
#' @slot type Model type, "binary" or "continuous"
#' @slot model_labels Label names and basic info for each model in the assessment
#' @slot auc_polygon_df Dataframe for plotting the AUC polygon
#' @slot roc_ci_df Dataframe for plotting the ROC confidence intervals
#' @slot roc_curve_df Dataframe for plotting the ROC curves
#' @slot auc_ci_labels_df Dataframe for AUC labels
#' @slot auc_tests Dataframe of significance tests for AUCs
#' @slot cat_data Dataframe of predictors (word engrams) with their model weights
#'
#' @export modelAssessment
#' @exportClass modelAssessment
#'
modelAssessment = setClass("modelAssessment", slots = c("type", "model_labels", "auc_polygon_df", "roc_ci_df", "roc_curve_df", "auc_ci_labels_df", "auc_tests", "cat_data"))
#' @title Create Model Assessment
#'
#' @description This function is deprecated; all dependent functions (e.g. \code{plot_roc()} or \code{plot_predictor_words}) now take individual models as arguments. This function assesses one or more models created by the \code{\link{language_model}} function.
#'
#' @param ... Models generated by the \code{\link{language_model}} function, and/or two-column dataframes with a predictor variable and an outcome variable
#'
#' @return An object of the type "modelAssessment"
#'
#' @import pROC
#' @importFrom methods setClass new
#' @importFrom rlang .data
#'
#' @export
#'
#' @examples
#'
#' \dontrun{
#' strong_movie_review_data$cleanText = clean_text(strong_movie_review_data$text)
#' mild_movie_review_data$cleanText = clean_text(mild_movie_review_data$text)
#'
#' # Using language to predict "Positive" vs. "Negative" reviews
#' # Only for strong reviews (ratings of 1 or 10)
#' movie_model_strong = language_model(strong_movie_review_data,
#' outcomeVariableColumnName = "valence",
#' outcomeVariableType = "binary",
#' textColumnName = "cleanText")
#'
#' # Using language to predict "Positive" vs. "Negative" reviews
#' # Only for mild reviews (ratings of 4 or 7)
#' movie_model_mild = language_model(mild_movie_review_data,
#' outcomeVariableColumnName = "valence",
#' outcomeVariableType = "binary",
#' textColumnName = "cleanText")
#'
#' # Create the model assessment
#' # movie_assessment = assess_models(movie_model_strong, movie_model_mild)
#' }
#'
#' @details
#' The primary purpose of this function is to be used with other functions included in this package, such as plot_roc() or predictor_word_plots()
#' All necessary calculations are performed by this function, so output plots and analyses can be performed quickly and modified as needed
#' This function can be used to assess models generated by the \code{\link{language_model}}, as well as simple predictors that could be compared with language models.
assess_models = function(...) {
dots = list(...)
dots_names = match.call(expand.dots = FALSE)
model_type = c()
for (i in 1:length(dots)) {
input = dots[[i]]
if (class(input) == "langModel") {
model_type = c(model_type, input@type)
}
else if (is.data.frame(input)) {
if (ncol(input) != 2) {
stop("Your dataframe '", as.character(dots_names$...[[i]]), "' should have 2 columns: a column of predictor variables, and a column of outcome variables.")
}
column_names = colnames(input)
if (is.factor(input[[column_names[2]]]) & nlevels(input[[column_names[2]]]) == 2) {
model_type = c(model_type, "binary")
}
else if (is.numeric(input[[column_names[2]]])) {
model_type = c(model_type, "continuous")
}
else {
stop(paste("The second (outcome) column of your dataframe '", as.character(dots_names$...[[i]]), "' must be a 2-level factor or numeric variable.", sep=""))
}
}
else {
stop(paste("Your argument '", as.character(dots_names$...[[i]]), "'must be either a langModel generated by the create_lasso_regression() function, or a two-column dataframe.", sep=""))
}
}
if (length(unique(model_type)) != 1) {
stop("All models being assessed must predict the same type of variable (binary outcomes or continuous outcomes).")
}
else {
model_type = model_type[1]
}
if (model_type == "binary") {
results_matrix = data.frame(matrix(nrow=0, ncol=8))
colnames(results_matrix) = c("model", "auc", "auc_ci", "predictive_acc", "high_predictor_1", "high_predictor_2", "low_predictor_1", "low_predictor_2")
model_labels = data.frame(matrix(ncol=7,nrow=0))
colnames(model_labels) = c("name","LASSO_model", "auc", "ci_lower", "ci_upper", "cat0title", "cat1title")
roc_plot_data = data.frame(matrix(ncol=2,nrow=0))
colnames(roc_plot_data) = c("specificities", "sensitivities")
roc_ci_plot_data = data.frame(matrix(ncol=4, nrow=0))
colnames(roc_ci_plot_data) = c("percent2p5", "percent50", "percent97p5", "sensitivities")
roc_list = list()
cat_data = data.frame(matrix(ncol=4,nrow=0))
colnames(cat_data) = c("words", "weights", "class", "model")
for (i in 1:length(dots)) {
input = dots[[i]]
if (class(input) == "langModel") {
LASSO_model = TRUE
predictor<- as.numeric(predict(input@cv,newx=input@x,s=input@lambda))
roc_data <- suppressMessages(roc(input@y,predictor, ci=TRUE))
roc_list[[as.character(dots_names$...[[i]])]] = roc_data
roc_ci_data = ci.sp(roc_data, sensitivities=seq(0,1,.01))
if (input@type == "binary") {
cat1title = paste('Words predicting\n"',input@level1, '" responses', sep="")
cat0title = paste('Words predicting\n"',input@level0, '" responses', sep="")
}
else {
cat1title = paste('Words predicting\n"high-',input@outcome, '" responses', sep="")
cat0title = paste('Words predicting\n"low-',input@outcome, '" responses', sep="")
}
cat0data = input@cat0raw
cat1data = input@cat1raw
if (nrow(cat0data) == 0) {
cat0data = data.frame(words="", weights=0)
}
if (nrow(cat1data) == 0) {
cat1data = data.frame(words="", weights=0)
}
cat0data$class = "cat0"
cat1data$class = "cat1"
cat_data_temp = rbind(cat0data, cat1data)
cat_data_temp$model = as.character(dots_names$...[[i]])
cat_data = rbind(cat_data, cat_data_temp)
}
else {
LASSO_model = FALSE
column_names = colnames(input)
input1_x = input[[column_names[1]]]
response = input[[column_names[2]]]
predictor_title1 = paste("'",column_names[1],"'", sep="")
response_title1 = paste("'",column_names[2],"'", sep="")
glm1<-glm(response~input1_x,family="binomial")
predictor = as.numeric(predict(glm1,newx=input1_x))
roc_data <- suppressMessages(roc(response,predictor, ci=TRUE))
roc_list[[as.character(dots_names$...[[i]])]] = roc_data
roc_ci_data = ci.sp(roc_data, sensitivities=seq(0,1,.01))
cat1title = ""
cat0title = ""
cat_data_temp = data.frame(words="", weights = 0, class=NA, model = as.character(dots_names$...[[i]]))
cat_data = rbind(cat_data, cat_data_temp)
}
roc_data_formatted = data.frame(specificities=roc_data$specificities, sensitivities=roc_data$sensitivities)
roc_data_formatted$model = as.character(dots_names$...[[i]])
roc_ci_data_formatted = as.data.frame(roc_ci_data)
roc_ci_data_formatted$sensitivities = as.numeric(rownames(roc_ci_data_formatted))
colnames(roc_ci_data_formatted) = c("percent2p5", "percent50", "percent97p5", "sensitivities")
roc_ci_data_formatted$model = as.character(dots_names$...[[i]])
roc_plot_data = rbind(roc_plot_data, roc_data_formatted)
roc_ci_plot_data = rbind(roc_ci_plot_data, roc_ci_data_formatted)
temp_frame = data.frame(name=as.character(dots_names$...[[i]]), LASSO_model=LASSO_model, auc = roc_data$auc, ci_lower = roc_data$ci[1], ci_upper = roc_data$ci[3], cat0title = cat0title, cat1title=cat1title)
model_labels = rbind(model_labels, temp_frame)
}
roc_plot_data$model = factor(roc_plot_data$model, levels = model_labels$name)
roc_ci_plot_data$model = factor(roc_ci_plot_data$model, levels = model_labels$name)
lowest_auc_model = model_labels$name[which.min(model_labels$auc)]
box_df = data.frame(polygon.x = c(0,0,1,1,0), polygon.y = c(0,1,1,0,0))
auc_df = data.frame(matrix(ncol=3,nrow=0))
colnames(auc_df) = c("sensitivities", "specificities", "model")
for (j in 1:nrow(model_labels)) {
auc_df_add = subset(roc_plot_data, model == model_labels$name[j])
auc_extra = data.frame(sensitivities = c(1,0,0), specificities = c(0,0,1), model=rep(model_labels$name[j],3))
auc_df = rbind(auc_df, auc_df_add, auc_extra)
}
auc_labels = data.frame(matrix(ncol=4,nrow=0))
colnames(auc_labels) = c("model", "label_text", "x", "y")
for (j in 1:nrow(model_labels)) {
model = model_labels$name[j]
label_text = paste("AUC: ",round(model_labels$auc[j],3), "\nCI: (", round(model_labels$ci_lower[j],3), "-", round(model_labels$ci_upper[j],3),")",sep="")
temp_frame = data.frame(model=model, label_text=label_text, x=.4, y=.6-(.1*j))
auc_labels = rbind(auc_labels, temp_frame)
}
roc_plot_data_all = CreateAllFacet(roc_plot_data, "model")
roc_ci_plot_data_all = CreateAllFacet(roc_ci_plot_data, "model")
auc_df$facet = auc_df$model
auc_df_add = subset(roc_plot_data, model == lowest_auc_model)
auc_df_add$facet = "all"
auc_extra = data.frame(sensitivities = c(1,0,0), specificities = c(0,0,1), model=rep(model_labels$name[j],3), facet=rep("all",3))
auc_df = rbind(auc_df, auc_df_add, auc_extra)
roc_plot_data_all$facet = factor(roc_plot_data_all$facet, levels = c(model_labels$name, "all"))
roc_ci_plot_data_all$facet = factor(roc_ci_plot_data_all$facet, levels = c(model_labels$name, "all"))
auc_df$facet = factor(auc_df$facet, levels = c(model_labels$name, "all"))
auc_labels$facet = "all"
auc_labels_new = auc_labels
auc_labels_new$facet = auc_labels_new$model
auc_labels_new$y=max(auc_labels_new$y)
auc_labels = rbind(auc_labels_new, auc_labels)
auc_labels$facet = factor(auc_labels$facet, levels = c(model_labels$name, "all"))
auc_tests = data.frame(matrix(ncol=5,nrow=0))
colnames(auc_tests) = c("model1", "model2", "model1_auc", "model2_auc", "p_value")
for (i in 1:length(roc_list)) {
for (j in (i+1):length(roc_list)) {
if (j > length(roc_list)) {
next
}
model1 = names(roc_list)[i]
model2 = names(roc_list)[j]
if (model1 == model2) {
next
}
model1_auc = roc_list[[i]]$auc
model2_auc = roc_list[[j]]$auc
p_value = roc.test(roc_list[[i]], roc_list[[j]])
temp_frame = data.frame(model1=model1, model2=model2, model1_auc=model1_auc, model2_auc=model2_auc, p_value=p_value$p.value)
auc_tests = rbind(auc_tests, temp_frame)
}
}
auc_tests$sig_TF = ifelse(auc_tests$p_value < .05, 1, 0)
auc_tests$sig_TF = factor(auc_tests$sig_TF, levels = c(0,1))
auc_tests$font = ifelse(auc_tests$sig_TF == 0, "plain", "bold")
auc_tests$size = ifelse(auc_tests$sig_TF == 0, 0, 1)
auc_tests$size = as.factor(auc_tests$size)
auc_tests$width = ifelse(auc_tests$sig_TF == 0, 1, .95)
auc_tests$height = ifelse(auc_tests$sig_TF == 0, 1, .95)
auc_tests$sig = ifelse(auc_tests$p_value >= .05, "NS", ifelse(auc_tests$p_value >= .01, "*", ifelse(auc_tests$p_value >= .001, "**", "***")))
auc_tests$model1 = factor(auc_tests$model1, levels = model_labels$name)
auc_tests$model2 = factor(auc_tests$model2, levels = rev(model_labels$name))
output = new("modelAssessment", type="binary", model_labels=model_labels, auc_polygon_df=auc_df, roc_ci_df=roc_ci_plot_data_all, roc_curve_df=roc_plot_data_all, auc_ci_labels_df=auc_labels, auc_tests=auc_tests, cat_data=cat_data)
return(output)
}
}
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.