Multiple Y variables, recoding continuous variables to categoriesand Medina/PI:

This first section will illustrate how to use more than one yvariable and how to generate a Median and a Ribbon showing a 95%Prediction interval (default) over the x variable (Time).

Using the built-in demo dataset:

• Change the mapped y variable(s) to Conc and Dose (ProTip: you candrag and drop the y variables to change their order)
• In theCategorize/cut subtab select Age to be cutinto three binned categories
• In theCategorize/cut subtab select Weight to becut into three quantiles categories
• Go toColor/Group/Split/Size/Fill Mappings and mapColour By and Fill By to Age and Column Split to Weight the Extra Rowsplit is automatically set to yvars since you have selected more thanone y variable.

cut a continuous variable to categorical

• Go toMedian PIs and select Median/PI and then youthis get this plot:

MedianPI

We can see that Dose does not change over time and that the highestAge category is only present in the second and third weight categories(older subjects have higher weights).

Boxplots, Median/PI, Mean:

• Remove all previous mappings from column split, colour by etc.
• Change the mapped y variable(s) to Weight and x variable to Age andremove Weight from Recode into Quantile Categories and select Ageinstead.
• Go toPoints, Lines and increase Point Size to 6and make the transparency of the points equal to 0.1
• Explore the jitter position including the custom one
• Go toColor/Group/Split/Size/Fill and map ColorBy:, Group By: Fill By: and Column Split: to Gender
• Go toOne Row by ID(s) and select ID so we keep onerow by ID

MedianPI

• Go ToMedian PI and uncheckIgnore MappedGroup so the Median PI uses the mapped Gender Group By:.
• Try to experiment what Label Values? and Label N? do Keep Label N?checked.
• Apply all the selected options in the screenshot

MedianPI

• Go toBoxplots and check the Add a Boxplot?checkbox.
• Go back toMedian PI and choose the PI to be 50% soit will be at the boxplot box edges.
• Go toBoxplots try to change the size of outliersand to remove the boxplot legend.

Boxplots

• Next go to theMean (CI) menu select Mean and checkShow points and Force Mean(s) Shape (Diamond will be used bydefault)
• Try to play with the various shapes options and or the size of themean point(s).
• Check Label Values? and Check Ignore Mapped Color and choose MeanLabel Geom as auto text repel.

MEANDIAMOND

Continuous and categorical variables descriptive stats:

In the following part we will generate a descriptive stats table thatreflect the plot that we just did and then add Race.

• Click on theDescriptive Stats Tab
• Map Extra Column Split to Gender and explore with the Flip the orderof the columns checkbox
• Try to add more statistics in the Statistics to display forcontinuous variables
• Add Race to the y variable(s) to see the statistics for acategorical variable

DescStats

Univariate Plots:

Remove all y variable(s) and any column splits keeping Age as xvariable gives a barplot since Age has been categorized.

barplots

Remove Age from Recode into Quantile Categories so it goes back to anumeric variable and the generated distribution will be a density plotinstead of a barplot. Reapply the ID inOne Row byID(s) as the data manipulation steps are sequential andchanging something in the first tab will reset the steps in thesubsequent ones.

distribution

Play with the options in theHistograms/Density/Barto see how they affect the generated plots.