Merge branch 'main' of https://github.com/easystats/performance

DESCRIPTION

@@ -1,7 +1,7 @@
 Type: Package
 Package: performance
 Title: Assessment of Regression Models Performance
-Version: 0.10.8.1
+Version: 0.10.8.2
 Authors@R: 
     c(person(given = "Daniel",
              family = "Lüdecke",

NAMESPACE

@@ -407,6 +407,7 @@ S3method(r2_coxsnell,coxph)
 S3method(r2_coxsnell,cpglm)
 S3method(r2_coxsnell,crch)
 S3method(r2_coxsnell,glm)
+S3method(r2_coxsnell,glmmTMB)
 S3method(r2_coxsnell,glmx)
 S3method(r2_coxsnell,logitmfx)
 S3method(r2_coxsnell,logitor)
@@ -463,6 +464,7 @@ S3method(r2_nagelkerke,coxph)
 S3method(r2_nagelkerke,cpglm)
 S3method(r2_nagelkerke,crch)
 S3method(r2_nagelkerke,glm)
+S3method(r2_nagelkerke,glmmTMB)
 S3method(r2_nagelkerke,glmx)
 S3method(r2_nagelkerke,logitmfx)
 S3method(r2_nagelkerke,logitor)

NEWS.md

@@ -1,3 +1,8 @@
+# performance 0.10.9
+
+* `r2()` for models of class `glmmTMB` without random effects now returns the
+  correct r-squared value for non-mixed models.
+
 # performance 0.10.8
 
 ## Changes

R/check_clusterstructure.R

@@ -47,14 +47,14 @@ check_clusterstructure <- function(x,
 
   H <- .clusterstructure_hopkins(x, distance = distance)
   if (H < 0.5) {
-    text <- paste0(
+    res_text <- paste0(
       "The dataset is suitable for clustering (Hopkins' H = ",
       insight::format_value(H),
       ").\n"
     )
     color <- "green"
   } else {
-    text <- paste0(
+    res_text <- paste0(
       "The dataset is not suitable for clustering (Hopkins' H = ",
       insight::format_value(H),
       ").\n"
@@ -67,7 +67,7 @@ check_clusterstructure <- function(x,
     dissimilarity_matrix = .clusterstructure_dm(x, distance = distance, method = "ward.D2")
   )
 
-  attr(out, "text") <- text
+  attr(out, "text") <- res_text
   attr(out, "color") <- color
   attr(out, "title") <- "Clustering tendency"
   class(out) <- c("see_check_clusterstructure", "check_clusterstructure", "easystats_check", class(out))
@@ -107,35 +107,32 @@ plot.check_clusterstructure <- function(x, ...) {
 
   n <- nrow(x) - 1
 
-  c <- apply(x, 2, min) # minimum value per column
+  cc <- apply(x, 2, min) # minimum value per column
   d <- apply(x, 2, max)
   p <- matrix(0, ncol = ncol(x), nrow = n) # n vectors of space
   for (i in seq_len(ncol(x))) {
-    p[, i] <- stats::runif(n, min = c[i], max = d[i])
+    p[, i] <- stats::runif(n, min = cc[i], max = d[i])
   }
   k <- round(stats::runif(n, 1, nrow(x)))
-  q <- as.matrix(x[k, ])
+  qq <- as.matrix(x[k, ])
   distp <- rep(0, nrow(x))
-  # distq=rep(0,nrow(x)-1)
   distq <- 0
   minp <- rep(0, n)
   minq <- rep(0, n)
   for (i in 1:n) {
     distp[1] <- stats::dist(rbind(p[i, ], x[1, ]), method = distance)
-    minqi <- stats::dist(rbind(q[i, ], x[1, ]), method = distance)
+    minqi <- stats::dist(rbind(qq[i, ], x[1, ]), method = distance)
     for (j in 2:nrow(x)) {
       distp[j] <- stats::dist(rbind(p[i, ], x[j, ]), method = distance)
-      error <- q[i, ] - x[j, ]
+      error <- qq[i, ] - x[j, ]
       if (sum(abs(error)) != 0) {
-        # distq[j]<-stats::dist(rbind(q[i,],x[j,]))
-        distq <- stats::dist(rbind(q[i, ], x[j, ]), method = distance)
+        distq <- stats::dist(rbind(qq[i, ], x[j, ]), method = distance)
         if (distq < minqi) {
           minqi <- distq
         }
       }
     }
     minp[i] <- min(distp)
-    # minq[i]<-apply(distq,1,min)
     minq[i] <- minqi
   }
   sum(minq) / (sum(minp) + sum(minq))

R/check_collinearity.R

@@ -405,10 +405,10 @@ check_collinearity.zerocount <- function(x,
 
 .check_collinearity <- function(x, component, ci = 0.95, verbose = TRUE) {
   v <- insight::get_varcov(x, component = component, verbose = FALSE)
-  assign <- .term_assignments(x, component, verbose = verbose)
+  term_assign <- .term_assignments(x, component, verbose = verbose)
 
   # any assignment found?
-  if (is.null(assign) || all(is.na(assign))) {
+  if (is.null(term_assign) || all(is.na(term_assign))) {
     if (verbose) {
       insight::format_alert(
         sprintf("Could not extract model terms for the %s component of the model.", component)
@@ -420,7 +420,7 @@ check_collinearity.zerocount <- function(x,
 
   # we have rank-deficiency here. remove NA columns from assignment
   if (isTRUE(attributes(v)$rank_deficient) && !is.null(attributes(v)$na_columns_index)) {
-    assign <- assign[-attributes(v)$na_columns_index]
+    term_assign <- term_assign[-attributes(v)$na_columns_index]
     if (isTRUE(verbose)) {
       insight::format_alert(
         "Model matrix is rank deficient. VIFs may not be sensible."
@@ -431,7 +431,7 @@ check_collinearity.zerocount <- function(x,
   # check for missing intercept
   if (insight::has_intercept(x)) {
     v <- v[-1, -1]
-    assign <- assign[-1]
+    term_assign <- term_assign[-1]
   } else {
     if (isTRUE(verbose)) {
       insight::format_alert("Model has no intercept. VIFs may not be sensible.")
@@ -443,22 +443,22 @@ check_collinearity.zerocount <- function(x,
   # hurdle or zeroinfl model can have no zero-inflation formula, in which case
   # we have the same formula as for conditional formula part
   if (inherits(x, c("hurdle", "zeroinfl", "zerocount")) &&
-    component == "zero_inflated" &&
-    is.null(f[["zero_inflated"]])) {
+      component == "zero_inflated" &&
+      is.null(f[["zero_inflated"]])) {
     f$zero_inflated <- f$conditional
   }
 
   if (inherits(x, "mixor")) {
-    terms <- labels(x$terms)
+    model_terms <- labels(x$terms)
   } else {
-    terms <- labels(stats::terms(f[[component]]))
+    model_terms <- labels(stats::terms(f[[component]]))
   }
 
   if ("instruments" %in% names(f)) {
-    terms <- unique(c(terms, labels(stats::terms(f[["instruments"]]))))
+    model_terms <- unique(c(model_terms, labels(stats::terms(f[["instruments"]]))))
   }
 
-  n.terms <- length(terms)
+  n.terms <- length(model_terms)
 
   if (n.terms < 2) {
     if (isTRUE(verbose)) {
@@ -476,7 +476,7 @@ check_collinearity.zerocount <- function(x,
   na_terms <- vector("numeric")
 
   for (term in 1:n.terms) {
-    subs <- which(assign == term)
+    subs <- which(term_assign == term)
     if (length(subs)) {
       result <- c(
         result,
@@ -489,7 +489,7 @@ check_collinearity.zerocount <- function(x,
 
   # any terms to remove, due to rank deficiency?
   if (length(na_terms)) {
-    terms <- terms[-na_terms]
+    model_terms <- model_terms[-na_terms]
   }
 
   # check for interactions, VIF might be inflated...
@@ -522,7 +522,7 @@ check_collinearity.zerocount <- function(x,
 
   out <- insight::text_remove_backticks(
     data.frame(
-      Term = terms,
+      Term = model_terms,
       VIF = result,
       VIF_CI_low = 1 / (1 - ci_lo),
       VIF_CI_high = 1 / (1 - ci_up),
@@ -538,7 +538,7 @@ check_collinearity.zerocount <- function(x,
 
   attr(out, "data") <- insight::text_remove_backticks(
     data.frame(
-      Term = terms,
+      Term = model_terms,
       VIF = result,
       SE_factor = sqrt(result),
       stringsAsFactors = FALSE
@@ -564,29 +564,29 @@ check_collinearity.zerocount <- function(x,
   tryCatch(
     {
       if (inherits(x, c("hurdle", "zeroinfl", "zerocount"))) {
-        assign <- switch(component,
+        term_assign <- switch(component,
           conditional = attr(insight::get_modelmatrix(x, model = "count"), "assign"),
           zero_inflated = attr(insight::get_modelmatrix(x, model = "zero"), "assign")
         )
       } else if (inherits(x, "glmmTMB")) {
-        assign <- switch(component,
+        term_assign <- switch(component,
           conditional = attr(insight::get_modelmatrix(x), "assign"),
           zero_inflated = .zi_term_assignment(x, component, verbose = verbose)
         )
       } else if (inherits(x, "MixMod")) {
-        assign <- switch(component,
+        term_assign <- switch(component,
           conditional = attr(insight::get_modelmatrix(x, type = "fixed"), "assign"),
           zero_inflated = attr(insight::get_modelmatrix(x, type = "zi_fixed"), "assign")
         )
       } else {
-        assign <- attr(insight::get_modelmatrix(x), "assign")
+        term_assign <- attr(insight::get_modelmatrix(x), "assign")
       }
 
-      if (is.null(assign)) {
-        assign <- .find_term_assignment(x, component, verbose = verbose)
+      if (is.null(term_assign)) {
+        term_assign <- .find_term_assignment(x, component, verbose = verbose)
       }
 
-      assign
+      term_assign
     },
     error = function(e) {
       .find_term_assignment(x, component, verbose = verbose)

R/check_distribution.R

@@ -77,7 +77,6 @@ check_distribution.default <- function(model) {
   } else {
     x <- stats::residuals(model)
   }
-  # x_scaled <- .normalize(x)
   dat <- .extract_features(x)
 
   dist_residuals <- as.data.frame(t(stats::predict(classify_distribution, dat, type = "prob")))
@@ -171,11 +170,11 @@ check_distribution.numeric <- function(model) {
   insight::check_if_installed("randomForest")
 
   dat <- .extract_features(model)
-  dist <- as.data.frame(t(stats::predict(classify_distribution, dat, type = "prob")))
+  distance <- as.data.frame(t(stats::predict(classify_distribution, dat, type = "prob")))
 
   out <- data.frame(
-    Distribution = rownames(dist),
-    p_Vector = dist[[1]],
+    Distribution = rownames(distance),
+    p_Vector = distance[[1]],
     stringsAsFactors = FALSE,
     row.names = NULL
   )

R/check_factorstructure.R

@@ -95,7 +95,7 @@ check_factorstructure <- function(x, n = NULL, ...) {
   kmo <- check_kmo(x, n, ...)
   sphericity <- check_sphericity_bartlett(x, n, ...)
 
-  text <- paste0("\n  - Sphericity: ", attributes(sphericity)$text, "\n  - KMO: ", attributes(kmo)$text)
+  res_text <- paste0("\n  - Sphericity: ", attributes(sphericity)$text, "\n  - KMO: ", attributes(kmo)$text)
 
   if (attributes(kmo)$color == "red" || attributes(sphericity)$color == "red") {
     color <- "red"
@@ -105,7 +105,7 @@ check_factorstructure <- function(x, n = NULL, ...) {
 
   out <- list(KMO = kmo, sphericity = sphericity)
 
-  attr(out, "text") <- text
+  attr(out, "text") <- res_text
   attr(out, "color") <- color
   attr(out, "title") <- "Is the data suitable for Factor Analysis?"
   class(out) <- c("easystats_check", class(out))
@@ -136,18 +136,16 @@ check_kmo <- function(x, n = NULL, ...) {
 
   # TODO: add interpret_kmo in effectsize and use that here for more fine-grained interpretation
   if (MSA < 0.5) {
-    text <-
-      sprintf(
-        "The Kaiser, Meyer, Olkin (KMO) overall measure of sampling adequacy suggests that factor analysis is likely to be inappropriate (KMO = %.2f).",
-        MSA
-      )
+    msg_text <- sprintf(
+      "The Kaiser, Meyer, Olkin (KMO) overall measure of sampling adequacy suggests that factor analysis is likely to be inappropriate (KMO = %.2f).", # nolint
+      MSA
+    )
     color <- "red"
   } else {
-    text <-
-      sprintf(
-        "The Kaiser, Meyer, Olkin (KMO) overall measure of sampling adequacy suggests that data seems appropriate for factor analysis (KMO = %.2f).",
-        MSA
-      )
+    msg_text <- sprintf(
+      "The Kaiser, Meyer, Olkin (KMO) overall measure of sampling adequacy suggests that data seems appropriate for factor analysis (KMO = %.2f).", # nolint
+      MSA
+    )
     color <- "green"
   }
 
@@ -160,9 +158,9 @@ check_kmo <- function(x, n = NULL, ...) {
     ")"
   ))
 
-  text <- paste0(text, " The individual KMO scores are: ", text_ind, ".")
+  msg_text <- paste0(msg_text, " The individual KMO scores are: ", text_ind, ".")
 
-  attr(out, "text") <- text
+  attr(out, "text") <- msg_text
   attr(out, "color") <- color
   attr(out, "title") <- "KMO Measure of Sampling Adequacy"
   class(out) <- c("easystats_check", class(out))
@@ -206,7 +204,7 @@ check_sphericity_bartlett <- function(x, n = NULL, ...) {
     color <- "red"
   }
 
-  attr(out, "text") <- text
+  attr(out, "text") <- msg_text
   attr(out, "color") <- color
   attr(out, "title") <- "Test of Sphericity"
   class(out) <- c("easystats_check", class(out))