Merge pull request #48 from dustinfife/development

Merge dev into master post plot.types

.Rbuildignore

@@ -17,15 +17,16 @@ README_files/
 tests/
 ^doc$
 ^Meta$
+jasp_documentation/
+paranormal.csv
 EJ notes.rtf
 R/flexplota.b.R
 # stuff to ignore jasp when doing CMD check
-R/glinmod_jasp.R
-R/linmod_jasp.R
-R/flexplot_jasp2.R
-R/mixedmod_jasp.R
-R/jasp_common.R
-
+# R/glinmod_jasp.R
+# R/linmod_jasp.R
+# R/flexplot_jasp2.R
+# R/mixedmod_jasp.R
+# R/jasp_common.R
 R/flexplota.h.R
 R/glmbasic.b.R
 R/glmbasic.h.R

DESCRIPTION

@@ -1,7 +1,7 @@
 Package: flexplot
 Type: Package
 Title: Graphically Based Data Analysis Using Flexplot
-Version: 0.7.2
+Version: 0.7.4
 Author: Dustin Fife
 Maintainer: Dustin Fife <fife.dustin@gmail.com>
 Description: The ‘flexplot’ suite is a graphically-based set of tools for doing data analysis. flexplot() allows users to specify a formula and the software automatically chooses what sort of graphic to present. Analysis can be paired with visuals using the visualize() function, such that the software will choose an appropriate graphic to match an R model object.
@@ -10,7 +10,6 @@ Encoding: UTF-8
 LazyData: true
 Depends: stats
 Imports: cowplot, 
-	ggplot2, 
 	MASS, 
 	tibble, 
 	withr, 
@@ -19,7 +18,8 @@ Imports: cowplot,
 	forcats, 
 	purrr,
 	plyr,
-	R6
+	R6,
+	ggplot2
 Suggests: jmvcore (>= 0.8.5), 
 	vdiffr,
 	knitr,

R/compare.fits.R

@@ -207,7 +207,11 @@ compare.fits = function(formula, data, model1, model2=NULL, return.preds=F, repo
 	if (return.preds){
 		prediction.model
 	} else {
-
+	  ### for logistic, add one to the predictions
+	  if (model1.type == "glm") {
+	    if (family(model1)$link=="logit" & !is.numeric(data[,outcome[1]])){
+	      prediction.model$prediction = prediction.model$prediction + 1
+	  }}
 		flexplot(formula, data=data, prediction=prediction.model, suppress_smooth=T, se=F, ...)
 	}	
 

R/data_documentation.R

@@ -1,3 +1,21 @@
+#' Simulated Dataset About Experiences with the Paranormal
+#'
+#' A dataset containing combat attributes of almost 812 fighters in the final
+#' Avengers battle. This dataset illustrates several types of problems one might
+#' encounter for univariate variables, including bimodality, zero-inflated data,
+#' outliers, mixed up labels, etc.
+#'
+#' @format A data frame with 1000 rows and 7 variables: \describe{
+#'   \item{conviction}{Degree of conviction they have about the existence of the
+#'   paranormal} \item{fear}{How much they fear being kidnapped by aliens}
+#'   \item{time}{How much time they spend a year researching the paranormal}
+#'   \item{kidnapped}{Whether they've been kidnapped by aliens}
+#'   \item{experiences.type}{What type of experiences they have had with the
+#'   paranormal? Can be "eerie feeling," "presence watching", "saw UFO", "saw ghost", or "heard voice"} 
+#'   \item{income}{How much money they make}
+#'   \item{age}{Age of respondent}}
+"paranormal"
+
 #' Simulated Statistics on the Final Avengers Battle
 #'
 #' A dataset containing combat attributes of almost 812 

R/estimates.R

@@ -226,7 +226,7 @@ estimates.lm = function(object, mc=TRUE){
 	    )
 	  }
 	  ### this requires superassignment to work with JASP
-	  #dataset<<-object$model
+	  dataset<<-object$model
 	  all.terms = attr(terms(object), "term.labels")
 	  mc = t(sapply(1:length(all.terms), removed.one.at.a.time, terms=all.terms, object=object))
 	  mc = data.frame(cbind(all.terms,mc), stringsAsFactors = FALSE)

R/flexplot.R

@@ -31,6 +31,9 @@
 ##' Users can export the string, in case they want to modify the ggplot object
 ##' @param silent Should all messages be suppressed? Defaults to F.
 ##' @param third.eye Should the "third eye" be employed? The third eye will be implemented shortly. 
+##' @param plot.type This argument allows the user to control the type of plot used. Flexplot defaults to histograms (for univariate variables)
+##' but could also do qqplots (using "qq" as the argument) or density plots (using "density"). Also, the user can specify "boxplot" for boxplots and
+##' "violin" for violin plots. 
 ##' @author Dustin Fife
 ##' @import tibble ggplot2 R6
 ##' @export
@@ -91,7 +94,8 @@ flexplot = function(formula, data=NULL, related=F,
 		spread=c('quartiles', 'stdev', 'sterr'), jitter=NULL, raw.data=T,
 		sample=Inf, 
 		prediction = NULL, suppress_smooth=F, alpha=.99977, plot.string=F, silent=F,
-		third.eye=NULL){
+		third.eye=NULL,
+		plot.type = c("histogram")){
 
 	#data = exercise_data
 	##### use the following to debug flexplot
@@ -102,10 +106,11 @@ flexplot = function(formula, data=NULL, related=F,
 	#formula = formula(weight.loss~rewards+gender|income+motivation); data=d; 
 	#ghost.reference = list(income=90000)
 
+
   spread = match.arg(spread, c('quartiles', 'stdev', 'sterr'))
-
-  ### prepare the variables
 
+  ### prepare the variables
+
   varprep = flexplot_prep_variables(formula, data, 
                                     breaks = breaks, labels=labels, bins=bins,
                                     related=related,  
@@ -124,6 +129,12 @@ flexplot = function(formula, data=NULL, related=F,
   ##### make models into a factor if they supply predictions
 	if (!is.null(prediction)){
 		prediction$model = factor(prediction$model)
+
+		### make the levels consistent between prediction/data for axis 1
+		if (!is.numeric(data[[varprep$axis[1]]])){
+		  prediction[[varprep$axis[1]]] = factor(prediction[[varprep$axis[1]]], levels=levels(data[[varprep$axis[1]]]))
+		}
+
 		varprep$prediction = prediction
 	}
 
@@ -150,19 +161,20 @@ flexplot = function(formula, data=NULL, related=F,
 	### PLOT UNIVARIATE PLOTS
   bivariate = with(varprep, flexplot_bivariate_plot(formula = NULL, data=data, prediction = prediction, outcome=outcome, predictors=predictors, axis=axis,
                                                     related=related, alpha=alpha, jitter=jitter, suppress_smooth=suppress_smooth, 
-                                                    method=method, spread=spread))
+                                                    method=method, spread=spread, plot.type=plot.type))
     p = bivariate$p
     points = bivariate$points
     fitted = bivariate$fitted
-
+    
 	#### all the above should take care of ALL possible plots, but now we add paneling
+   # browser()
   facets = flexplot_panel_variables(varprep, 
                                     related=related, labels=labels, bins=bins, 
                                     suppress_smooth=suppress_smooth, method=method, 
                                     spread=spread, prediction=prediction)
 
 	if (!is.null(ghost.line) & !is.na(varprep$given[1])){ # with help from https://stackoverflow.com/questions/52682789/how-to-add-a-lowess-or-lm-line-to-an-existing-facet-grid/52683068#52683068
-	  
+
 				### bin the ghost reference if it's not null
     ghost.reference = with(varprep, create_ghost_reference(ghost.reference=ghost.reference, data=data,
                                              bins=bins, breaks=breaks, given=given, axis=axis, labels=labels))
@@ -184,7 +196,7 @@ flexplot = function(formula, data=NULL, related=F,
 	} else {
 		ghost = "xxxx"
 	}	
-                                                             
+
 	### add prediction lines
 	if (!is.null(prediction)){
 		### see how many models are being compared
@@ -216,7 +228,7 @@ flexplot = function(formula, data=NULL, related=F,
 		#### change the y axis labels
 		theme = paste0(theme, " + scale_y_continuous(breaks = c(0,1), labels=factor.to.logistic(data,outcome, labels=T))")	
 	}
-	
+
 	### put objects in this environment
 	axis = varprep$axis; outcome = varprep$outcome; predictors = varprep$predictors; levels = length(unique(data[,outcome]))