Baysig documentation: Baysig plots

Plots can be embedded in Baysig documents generated on BayesHive. To generate a plot, put a value of a Plot type in a question block, for instance:

 Here is my plot
 ?> distPlot (normal 0 1)
 Thanks for watching!

In the following sections, we describe the different ways to create and manipulate Plot values.

Fundamental plots

There are a number of functions that take data of different types and turn them into plots

Distribution histograms

distPlot takes a distribution over real numbers and returns a histogram plot.

The hitogram plots give you a small input control to set the number of bins. This can be dynamically changed by the user.

Data histograms

histogram takes a list of numbers (real or integers) as its argument and returns a histogram plot

fromTo 1 10 ⇒ [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]

Scatter plots

scatterPlot takes a list of pairs of numbers (either of which can be real or integers) and creates a scatter plot using the pairs as coordinates (as is conventional, the first pair component is plotted on the x-axis and the second on the y-axis).

xyData = zipWithNats (linspace 0 10 8) 1

xyData ⇒ [(1,0.0000), (2,1.2500), (3,2.5000), (4,3.7500), (5,5.0000), (6,6.2500), (7,7.5000), (8,8.7500)]

Signal plots

sigPlot plots a timeseries from a value of type Real->Real.

We Solve a simple differential equation to generate data for this example, but you can also use uploaded timeseries.

dt = 0.1
tmax = 10
s_0 = 1.0
D s t = - s t

Probabilistic signals, points and lines

psigPlot is the probabilistic equivalent of sigPlot which takes a probability distribution over timeseries instead of a single value. Let's plot the wiener distribution over Wiener processes:

As you can see, it draws a number of samples from the probability distribution. You ca change the number of samples drawn by using psigNPlot instead:

If you plot a lot of signals this way, your browser will feel very slow.

ppoints is the probabilistic equivalent of scatterPlot. It takes a probability distribution over list of pairs. ppoints does the same thing, but draws lines between the points. We will a suitable value here to illustrate:

myRandomPoints = prob
   xs = linspace 0 1 10
   ys ~ mapM (\x -> normal x 0.1) xs
   return (zip xs ys)

This is pur first example of multi-panel plots, which will be described below.

Plot modifiers

There are a number of function that modify fundamental plots.

over superimposes a list of plots on a single plot with a single set of axes.

axisLabels sets the plot x axis label (first argument) and y axis label (second argument) on the plot which is its third argument

style allows you to specify the styles that correspond to Radian style properties. These are specified as a list of string-string pairs.

Style properties

See Radian documentation

Multipanel plots

The constructors PlotRow and PlotColumn can be used to nest plots. Both take an initial argument with is a list of style properties for that whole sub-plot.

Writing new plotting functions (advanced)

TODO