Wtih constant data fit(Histogram) sometimes gives a wrong result

[vincent-picaud]

In some circumstances, fit(Histogram) gives a wrong result.

An example that gives the expected result : weight = 5 in [0,1)

julia> using StatsBase

julia> d=zeros(5)
5-element Vector{Float64}:
 0.0
 0.0
 0.0
 0.0
 0.0

julia> StatsBase.fit(Histogram,d)
Histogram{Int64, 1, Tuple{StepRangeLen{Float64, Base.TwicePrecision{Float64}, Base.TwicePrecision{Float64}, Int64}}}
edges:
  0.0:1.0:1.0
weights: [5]
closed: left
isdensity: false

however, with another constant data value the result makes no sense (to me) : weight = 0 in [1.1, 3.1)

julia> d.+=0.1
5-element Vector{Float64}:
 0.1
 0.1
 0.1
 0.1
 0.1

julia> StatsBase.fit(Histogram,d)
Histogram{Int64, 1, Tuple{StepRangeLen{Float64, Base.TwicePrecision{Float64}, Base.TwicePrecision{Float64}, Int64}}}
edges:
  1.1:2.0:3.1
weights: [0]
closed: left
isdensity: false

My Julia version :

julia> versioninfo()
Julia Version 1.7.1
Commit ac5cc99908 (2021-12-22 19:35 UTC)
Platform Info:
  OS: Linux (x86_64-linux-gnu)
  CPU: Intel(R) Core(TM) i7-9850H CPU @ 2.60GHz
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-12.0.1 (ORCJIT, skylake)

[devmotion]

The issue is fixed on the master branch, probably by #792:

julia> ] add StatsBase#master

julia> using StatsBase

julia> d=zeros(5)
5-element Vector{Float64}:
 0.0
 0.0
 0.0
 0.0
 0.0

julia> StatsBase.fit(Histogram,d)
Histogram{Int64, 1, Tuple{StepRangeLen{Float64, Base.TwicePrecision{Float64}, Base.TwicePrecision{Float64}, Int64}}}
edges:
  0.0:1.0:1.0
weights: [5]
closed: left
isdensity: false

julia> d.+=0.1
5-element Vector{Float64}:
 0.1
 0.1
 0.1
 0.1
 0.1

julia> StatsBase.fit(Histogram,d)
Histogram{Int64, 1, Tuple{StepRangeLen{Float64, Base.TwicePrecision{Float64}, Base.TwicePrecision{Float64}, Int64}}}
edges:
  0.1:1.0:1.1
weights: [5]
closed: left
isdensity: false

[devmotion]

A new release should be available quite soon: #803

[vincent-picaud]

@devmotion thanks for the quick answer and sorry for not being aware of #803.

[vincent-picaud]

Should I close the issue?

[devmotion]

That depends on you 🙂 I think it's OK to keep it open until a new release is available. But if you think your problem is addressed sufficiently since it is fixed on the master branch, feel free to close it.

[vincent-picaud]

As the problem is fixed I will close it. Thanks