The distribution zoo

App to view distribution properties and access dynamic code in R, Python, Matlab, Mathematica and Stan. The app is available online here.

To add a new distribution

• Add distribution to relevant conditional panel at top of ui.R.
• Add slider inputs for the parameter values in ui.R and, if necessary, range parameters that will determine plotting range.
• Add the mean and variance in fCalculateMeanFull and fCalculateVarianceFull within functions.R.
• Add arguments in one of the fExtra[]FunctionInputsFull functions that will be passed to the function to call it (for plotting) to plotting.R. For example, the normal distribution is parameterised by two named arguments in R's implementation mean and sd. So to plot this distribution as we vary the user-selected parameters input$normal_mu and input$normal_sigma, we need to make a named argument call of the form paste("mean=",input$normal_mu,",sd=",input$normal_sigma), which is pasted than evaluated to plot.
• Add a scale for discrete distributions in fScaleFull1 or continuous distributions in fScaleFull in plotting.R.
• Add function to call to evaluate PDF/PMF in the data <- reactive({ function in server.R. Note, if this function does not exist in base R or within a package, you can define a custom one within the functions.R file. Note the named arguments to the function must the same as that specified in the above step which takes place in plotting.R.
• Add function to call to evaluate CDF in the dataCDF <- reactive({ function in server.R. Note, see above for custom functions.
• Add formulae in latex form in formulae.R.
• Copy code from formulae.R and put into correct form using helper functions (fLatexHelper_) where possible which aid with the construction of code that renders nicely in the app.
• Add code for R, Python, Mathematica, Matlab and Stan by updating the relevant code_ files. Note that the code is intended to be dynamic and the properties shown on the plots should be exactly replicated by the code examples. This means that the parameterisation in the function call may not reflect the default parameterisation.
• Add example likelihood and prior uses for the distribution in example_users.R.