Plot function #8
Conversation
Separetes plot functions into three functions 1. Default: Plots density ratio histograms of both samples 2. Univariate: Plot density ratio value against value of one variable. 3. Bivariate: Plot density ratio value (in colors), for combinations of two variables
implemented several changes in plot functions (separe objects, simply code, use base pipe, use default bins, separate bins visually, add argument to plot x with logscale, add argument for shapes)
checks in all plots for object type, names not overriden, names in original dataset
- Bivariate plot fuctions plots bivariate plots of chosen variables - Adjust some aesthethics (fill, colour, alpha, shape) to improve visibility - Add roxigeny2 to include documentation (+export)
Solve a couple of small bugs,added informative messages, added histogram binwidth as an argument
Remove duplicate variables, add heatmap
changed histogram function, change warning called, changed value transformation
Fix small bug related to variable names in variable grid. Make both bivariates plot show logarithm(density ratio) by default, and ajust scales based on it
small bug fix with heatmap related to negative predicted density ratios
added bins argument for histogram; made logarithmic transformation warning conditional on negative values present, mapped colours to sample in univariate plot
improved warning messages; make bivariate plot only show the lower diagonal
minor edit in colours bivariate heatmap
Solve small bug warning, rearrageged numerator and denominator labels univariate plot, create assembled outplot in univariate plots, implemented "smart" labels on y axis
fix labels, add possibility of choosing number of rows univariate plot
add show.sample argument for bivariate plots, add printing method
R/plot.R
Outdated
| #' | ||
| #' Plots a scatterplot of two variables, with densityratio mapped to the colour scale. | ||
| #' | ||
| #' @inheritParams plot.univariate |
There was a problem hiding this comment.
It did not work. The solution suggested here (https://stackoverflow.com/questions/47782004/how-to-use-inheritparams-on-single-parameters-when-multiple-parameters-match) involves creating additional files, which I think is more cumbersome than helpful for a single repeated parameter. So I just solved it by defining the repeated parameter twice.
.gitignore
Outdated
| @@ -2,3 +2,4 @@ | |||
| .Rhistory | |||
| .RData | |||
| .Ruserdata | |||
| \playground | |||
There was a problem hiding this comment.
Don't forget to remove the playground folder again (and also remove it from the .gitignore then).
R/plot.R
Outdated
| #' | ||
| #' @examples | ||
| plot_bivariate <- function(object, vars, samples = "both", | ||
| output = "assembled", logscale = TRUE, show.sample = FALSE) { |
There was a problem hiding this comment.
Can we also have vars1, vars2 = NULL, and if vars2 is specified, it must be of the same length as vars1, and specifies what each variable in vars1 must be plotted against (this can be nice if you have many variables and only want to look at some combinations). Then we can simply have (later on)
if (is.null(vars2)) {
var_combinations = expand.grid(vars1, vars1)
} else {
var_combinations = matrix(c(vars1, vars2), ncol = 2)
}
There was a problem hiding this comment.
Did that! I also added one check so that if the length of vars1 == 1 and vars2 == 1, then it defaults to grid = FALSE (A grid makes no sense for one plot). And another one so that if length(vars1) and &is.nul(vars2), then there's an error (youneed to variables for a bivariate plot).
R/plot.R
Outdated
|
|
||
| # Erase upper diagonal | ||
| ## Create plot into a grob | ||
| grob <- ggplotGrob(plot) |
There was a problem hiding this comment.
If we add the vars2 = NULL argument, we also need to adjust this code here, such that it is only executed if is.null(vars2).
There was a problem hiding this comment.
I don't think so. Var2 and Var1 are the defaults names created when you do expand.grid (maybe that's where the confusion comes from?). This code is just removing the upper diagonal. It is working right now regardless of whether you specify var2 or not.
CarlosPoses
force-pushed
the
plot-feature
branch
from
d013e84 to
ecbc903
Compare
CarlosPoses
force-pushed
the
plot-feature
branch
from
ecbc903 to
5de9b38
Compare
Solve final issues before CRAN submission. Concretely: - solve small typos/improve comments - creates function check.logscale() - adds plot.lhss and plot.spectral - minor changes to legiility of code - adds check.var.combinations function
Hi @thomvolker , here the first draft of the plot function. A lot of features to add and discuss yet.