Data.Histogram.Tutorial

1.

The first example illustrates one of the most common use-cases of a histogram, i.e.

• uniformly-spaced, equal-weight bins
• one-dimensional distribuded data
• binning range equal to the data range

We can write a helper function that populates a Histogram from a Foldable container (e.g. an array, or a Vector, or a tree, etc.) of Doubles :

histo :: (Foldable v, Unbox a, Num a) =>
         Int
      -> v Double
      -> Histogram BinD a
histo n v = fillBuilder buildr v
  where
    mi = minimum v
    ma = maximum v
    bins = binD mi n ma
    buildr = mkSimple bins

We can now declare our first histogram with 4 bins and a list of data :

> let h0 = histo 4 [1,2,3,5,1,-10,2,3,50,1,6,7,4,6,34,45,20,120,-80]

The Show instance of Histogram lets us see the histogram metadata :

> h0
# Histogram
# Underflows = 0.0
# Overflows  = 1.0
# BinD
# Base = -80.0
# Step = 50.0
# N    = 4
-55.0	1.0
-5.0	13.0
45.0	4.0
95.0	0.0

Note : with this binning algorithm, the bin intervals are closed to the left and open to the right, which is why the 120 element is marked as an overlow.

Note 2 : the output of show shouldn't generally be used as a form of data serialization.

The data bin centers and bin counts can be retrieved with asList:

> asList h0
[(-55.0,1.0),(-5.0,13.0),(45.0,4.0),(95.0,0.0)]