Skip to content

Commit

Permalink
use AbstractVector and include join method
Browse files Browse the repository at this point in the history
  • Loading branch information
@diegozea
diegozea committed Jan 26, 2016
1 parent b3c693e commit 0a85c61
Show file tree
Hide file tree
Showing 6 changed files with 204 additions and 43 deletions.
6 changes: 6 additions & 0 deletions NEWS.md
View file
Original file line number Diff line number Diff line change
@@ -1,5 +1,11 @@
## PairwiseListMatrices.jl Release Notes

### Changes from v0.2 to v0.3

* `join` two PairwiseListMatrices by their labels

* `PairwiseListMatrix` nows works with any `AbstractVector`. i.e. `DataArray`s

### Changes from v0.1 to v0.2

* `hasdiagonal` returns the `diagonal` boolean value of `PairwiseListMatrix{T, diagonal}`
Expand Down
34 changes: 17 additions & 17 deletions README.md
View file
Original file line number Diff line number Diff line change
@@ -1,29 +1,29 @@
# PairwiseListMatrices

Julia 0.4: [![PairwiseListMatrices](http://pkg.julialang.org/badges/PairwiseListMatrices_0.4.svg)](http://pkg.julialang.org/?pkg=PairwiseListMatrices)
Linux, OSX: [![Build Status](https://travis-ci.org/diegozea/PairwiseListMatrices.jl.svg?branch=master)](https://travis-ci.org/diegozea/PairwiseListMatrices.jl)
Windows: [![Build status](https://ci.appveyor.com/api/projects/status/p96sso5b23gi85mg/branch/master?svg=true)](https://ci.appveyor.com/project/diegozea/pairwiselistmatrices-jl/branch/master)
Julia 0.4: [![PairwiseListMatrices](http://pkg.julialang.org/badges/PairwiseListMatrices_0.4.svg)](http://pkg.julialang.org/?pkg=PairwiseListMatrices)
Linux, OSX: [![Build Status](https://travis-ci.org/diegozea/PairwiseListMatrices.jl.svg?branch=master)](https://travis-ci.org/diegozea/PairwiseListMatrices.jl)
Windows: [![Build status](https://ci.appveyor.com/api/projects/status/p96sso5b23gi85mg/branch/master?svg=true)](https://ci.appveyor.com/project/diegozea/pairwiselistmatrices-jl/branch/master)
Code Coverage: [![Coverage Status](https://coveralls.io/repos/diegozea/PairwiseListMatrices.jl/badge.svg?branch=master&service=github)](https://coveralls.io/github/diegozea/PairwiseListMatrices.jl?branch=master) [![codecov.io](http://codecov.io/github/diegozea/PairwiseListMatrices.jl/coverage.svg?branch=master)](http://codecov.io/github/diegozea/PairwiseListMatrices.jl?branch=master)

**Documentation:** [http://diegozea.github.io/PairwiseListMatrices.jl/](http://diegozea.github.io/PairwiseListMatrices.jl/)

## Description
## Description

This package allows you to use a pairwise list as a matrix:
![PLM](https://raw.githubusercontent.com/diegozea/PairwiseListMatrices.jl/master/doc/PLM_README.png)
This package allows you to use a pairwise list as a matrix:

![PLM](https://raw.githubusercontent.com/diegozea/PairwiseListMatrices.jl/master/doc/PLM_README.png)

```julia
type PairwiseListMatrix{T, diagonal} <: AbstractArray{T, 2}
list::Vector{T}
diag::Vector{T}
list::AbstractVector{T}
diag::AbstractVector{T}
labels::IndexedArray
nelements::Int
end
```
`PairwiseListMatrix{T, diagonal}` is a (squared) symmetric matrix that stores a `list` of values of type `T` for the pairwise comparison/measure of `nelements`.
If the parameter `diagonal` is `true` the first element of the list is `1, 1`, otherwise is `1, 2`. The diagonal values are stored in a vector on the `diag` field when `diagonal` is `false`.
Labels can be stored on the field `labels` as an [IndexedArray](https://github.com/garrison/IndexedArrays.jl).
`PairwiseListMatrix{T, diagonal}` is a (squared) symmetric matrix that stores a `list` of values of type `T` for the pairwise comparison/measure of `nelements`.
If the parameter `diagonal` is `true` the first element of the list is `1, 1`, otherwise is `1, 2`. The diagonal values are stored in a vector on the `diag` field when `diagonal` is `false`.
Labels can be stored on the field `labels` as an [IndexedArray](https://github.com/garrison/IndexedArrays.jl).

## Features

Expand All @@ -39,11 +39,11 @@ In pairwise calculations like `cor()` if results are saved as `PairwiseListMatri
- [Creation benchmark](http://nbviewer.ipython.org/github/diegozea/PairwiseListMatrices.jl/blob/master/test/creation_bech.ipynb)
- [Statistics benchmark](http://nbviewer.ipython.org/github/diegozea/PairwiseListMatrices.jl/blob/master/test/stats_bench.ipynb)

#### Plots
Since this could be a representation for an **adjacency matrix/list** of an undirected graph, the function `protovis` provides an **arc diagram** and a **matrix visualization** on the web browser using [Protovis](http://mbostock.github.io/protovis/).
![Protovis example](https://raw.githubusercontent.com/diegozea/PairwiseListMatrices.jl/master/doc/protovis_example.png)
#### Plots
Since this could be a representation for an **adjacency matrix/list** of an undirected graph, the function `protovis` provides an **arc diagram** and a **matrix visualization** on the web browser using [Protovis](http://mbostock.github.io/protovis/).

![Protovis example](https://raw.githubusercontent.com/diegozea/PairwiseListMatrices.jl/master/doc/protovis_example.png)

## Example

```julia
Expand Down
4 changes: 2 additions & 2 deletions doc/index.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -69,8 +69,8 @@ All `PairwiseListMatrix{T, diagonal}` have a `list` field for storing the values

```{.julia execute="false"}
type PairwiseListMatrix{T, diagonal} <: AbstractArray{T, 2}
list::Vector{T}
diag::Vector{T}
list::AbstractVector{T}
diag::AbstractVector{T}
labels::IndexedArray
nelements::Int
end
Expand Down
2 changes: 1 addition & 1 deletion src/PairwiseListMatrices.jl
View file
Original file line number Diff line number Diff line change
Expand Up @@ -8,7 +8,7 @@ module PairwiseListMatrices
import Base: size, getindex, setindex!, length, eltype,
similar, copy, abs, full, zeros, ones, convert,
transpose, transpose!, ctranspose, ctranspose!, diag, svd,
triu!, triu, writedlm, writecsv,
triu!, triu, writedlm, writecsv, join,
-, +, .*, .-, .+, .-, ./, *, /,
# faster
mean, sum, varm, var, std, sqrt
Expand Down
163 changes: 140 additions & 23 deletions src/pairwiselistmatrix.jl
View file
Original file line number Diff line number Diff line change
Expand Up @@ -5,8 +5,8 @@ If `diagonal` is `false` the diagonal values are stored in a vector on the `diag
Labels can be stored on the field `labels` as an `IndexedArray`.
"""
type PairwiseListMatrix{T, diagonal} <: AbstractArray{T, 2}
list::Vector{T}
diag::Vector{T}
list::AbstractVector{T} # i.e. It can be a DataArray
diag::AbstractVector{T}
labels::IndexedArray
nelements::Int
end
Expand Down Expand Up @@ -64,7 +64,7 @@ lengthlist(nelements::Int, diagonal::Bool) = lengthlist(nelements, Val{diagonal}
# =============

ij2k{diagonal}(i::Int, j::Int, nelements::Int, ::Type{Val{diagonal}}) = div(diagonal ? (nelements*(nelements+1))-((nelements-i)*(nelements-i+1)) :
(nelements*(nelements-1))-((nelements-i)*(nelements-i-1)), 2) - nelements + j
(nelements*(nelements-1))-((nelements-i)*(nelements-i-1)), 2) - nelements + j

"""Returns the `k` index of the `list` from the indixes `i` and `j` with `i<j` from a matrix of `nelements` by `nelements`.
`diagonal` should be `true` or `Val{true}` if the diagonal values are on the `list`. You must not use it with `i>j`.
Expand Down Expand Up @@ -97,7 +97,7 @@ If `diagonal` is `false` the diagonal values are stored in a vector on the `diag
This vector can be filled with the optional `diagonalvalue` arguments (default to `0`).
"""
function PairwiseListMatrix{T}(::Type{T}, nelements::Int, labels::IndexedArray,
diagonal::Bool=false, diagonalvalue::T=zero(T))
diagonal::Bool=false, diagonalvalue::T=zero(T))
_test_nelements(labels, nelements, "labels")
if diagonal
return PairwiseListMatrix{T, diagonal}(Array(T, lengthlist(nelements, Val{true})), T[], labels, nelements)
Expand All @@ -116,8 +116,8 @@ PairwiseListMatrix{T}(::Type{T}, nelements::Int, diagonal::Bool=false,
# -------

convert{T, F, D}(::Type{PairwiseListMatrix{T, D}},
mat::PairwiseListMatrix{F, D}) = PairwiseListMatrix{T, D}(convert(Vector{T}, mat.list),
convert(Vector{T}, mat.diag), copy(mat.labels), copy(mat.nelements))
mat::PairwiseListMatrix{F, D}) = PairwiseListMatrix{T, D}(convert(Vector{T}, mat.list),
convert(Vector{T}, mat.diag), copy(mat.labels), copy(mat.nelements))

function convert{T, F}(::Type{PairwiseListMatrix{T, true}}, mat::PairwiseListMatrix{F, false})
N = copy(mat.nelements)
Expand Down Expand Up @@ -181,7 +181,7 @@ function convert{T, F}(::Type{Array{F,2}}, lm::PairwiseListMatrix{T, true})
k = 0
for col in 1:N
@inbounds for row in col:N
complete[row, col] = complete[col, row] = F(list[k += 1])
complete[row, col] = complete[col, row] = F(list[k += 1])
end
end
complete
Expand Down Expand Up @@ -258,9 +258,9 @@ length(m::PairwiseListMatrix) = m.nelements * m.nelements
eltype{T, diagonal}(m::PairwiseListMatrix{T, diagonal}) = T

similar{T, diagonal}(m::PairwiseListMatrix{T, diagonal}) = PairwiseListMatrix{T, diagonal}(similar(m.list), copy(m.diag),
copy(m.labels), copy(m.nelements))
copy(m.labels), copy(m.nelements))
similar{T, diagonal, S}(m::PairwiseListMatrix{T, diagonal}, ::Type{S}) = PairwiseListMatrix{S, diagonal}(similar(m.list, S), similar(m.diag, S),
copy(m.labels), copy(m.nelements))
copy(m.labels), copy(m.nelements))

copy{T, diagonal}(m::PairwiseListMatrix{T, diagonal}) = PairwiseListMatrix{T, diagonal}(copy(m.list), copy(m.diag), copy(m.labels), copy(m.nelements))

Expand Down Expand Up @@ -461,11 +461,11 @@ for bin in (:/, :./)

@eval begin

$(bin){T <: AbstractFloat}(A::PairwiseListMatrix{T, true}, B::T) = PairwiseListMatrix{T, true }($(bin)(A.list, B), copy(A.diag), copy(A.labels), copy(A.nelements))
$(bin){T <: AbstractFloat}(A::PairwiseListMatrix{T, false}, B::T) = PairwiseListMatrix{T, false}($(bin)(A.list, B), $(bin)(A.diag, B), copy(A.labels), copy(A.nelements))
$(bin){T <: AbstractFloat}(A::PairwiseListMatrix{T, true}, B::T) = PairwiseListMatrix{T, true }($(bin)(A.list, B), copy(A.diag), copy(A.labels), copy(A.nelements))
$(bin){T <: AbstractFloat}(A::PairwiseListMatrix{T, false}, B::T) = PairwiseListMatrix{T, false}($(bin)(A.list, B), $(bin)(A.diag, B), copy(A.labels), copy(A.nelements))

$(bin){T}(A::PairwiseListMatrix{T, true}, B::Integer) = PairwiseListMatrix{Float64, true }($(bin)(A.list, B), copy(A.diag), copy(A.labels), copy(A.nelements))
$(bin){T}(A::PairwiseListMatrix{T, false}, B::Integer) = PairwiseListMatrix{Float64, false}($(bin)(A.list, B), $(bin)(A.diag, B), copy(A.labels), copy(A.nelements))
$(bin){T}(A::PairwiseListMatrix{T, true}, B::Integer) = PairwiseListMatrix{Float64, true }($(bin)(A.list, B), copy(A.diag), copy(A.labels), copy(A.nelements))
$(bin){T}(A::PairwiseListMatrix{T, false}, B::Integer) = PairwiseListMatrix{Float64, false}($(bin)(A.list, B), $(bin)(A.diag, B), copy(A.labels), copy(A.nelements))

end

Expand All @@ -482,8 +482,8 @@ for bin in (:.+, :.-, :.*, :-, :+)

@eval begin

$(bin){T <: Number}(A::PairwiseListMatrix{T, true}, B::T) = PairwiseListMatrix{T, true }($(bin)(A.list, B), copy(A.diag), copy(A.labels), copy(A.nelements))
$(bin){T <: Number}(A::PairwiseListMatrix{T, false}, B::T) = PairwiseListMatrix{T, false}($(bin)(A.list, B), $(bin)(A.diag, B), copy(A.labels), copy(A.nelements))
$(bin){T <: Number}(A::PairwiseListMatrix{T, true}, B::T) = PairwiseListMatrix{T, true }($(bin)(A.list, B), copy(A.diag), copy(A.labels), copy(A.nelements))
$(bin){T <: Number}(A::PairwiseListMatrix{T, false}, B::T) = PairwiseListMatrix{T, false}($(bin)(A.list, B), $(bin)(A.diag, B), copy(A.labels), copy(A.nelements))

end

Expand Down Expand Up @@ -675,20 +675,20 @@ function varm{T <: PairwiseListMatrix}(list::Vector{T}, mean::PairwiseListMatrix
end
end
if !_has_diagonal(out)
out_diag = out.diag
mean_diag = mean.diag
@inbounds for sample in list
sample_diag = sample.diag
for k in 1:length(out_diag)
@inbounds out_diag[k] += abs2( sample_diag[k] - mean_diag[k])
out_diag = out.diag
mean_diag = mean.diag
@inbounds for sample in list
sample_diag = sample.diag
for k in 1:length(out_diag)
@inbounds out_diag[k] += abs2( sample_diag[k] - mean_diag[k])
end
end
end
end
out ./ samples
end

function var{T <: PairwiseListMatrix}(list::Vector{T}; mean=nothing)
mean === nothing ? varm(list, Base.mean(list)) : varm(list, mean)
mean === nothing ? varm(list, Base.mean(list)) : varm(list, mean)
end

std{T <: PairwiseListMatrix}(list::Vector{T}; mean=nothing) = sqrt(var(list, mean=mean))
Expand Down Expand Up @@ -892,3 +892,120 @@ writecsv(filename::ByteString, plm::PairwiseListMatrix, diagonal::Bool=true) = w
triu!(::PairwiseListMatrix, args...) = throw(ErrorException("PairwiseListMatrix must be Symmetric, use triu instead of triu!"))
triu(mat::PairwiseListMatrix, args...) = triu(full(mat), args...)

# join
# ====

"""
This function join two PairwiseListMatrices by their labels, returning two PairwiseListMatrices with same size and labels.
There are 4 `kind` of joins:
- `:inner` : Intersect. The output matrices only include the labels that are in both PairwiseListMatrices
- `:outer` : Union. Include the labels of the two PairwiseListMatrices.
- `:left` : Only use labels form the first argument.
- `:right` : Only use labels form the second argument.
`NaN`s are filled in where needed to complete joins. The default value for missing values can be changed passing a tuple to `missing`.
```
julia> l = PairwiseListMatrix([1.,2.,3.], ['a', 'b', 'c'], false) # a b c
3x3 PairwiseListMatrices.PairwiseListMatrix{Float64,false}:
0.0 1.0 2.0
1.0 0.0 3.0
2.0 3.0 0.0
julia> r = PairwiseListMatrix([1.,2.,3.], ['b', 'c', 'd'], false) # b c d
3x3 PairwiseListMatrices.PairwiseListMatrix{Float64,false}:
0.0 1.0 2.0
1.0 0.0 3.0
2.0 3.0 0.0
julia> join(l, r, kind=:inner) # b c
(
2x2 PairwiseListMatrices.PairwiseListMatrix{Float64,false}:
0.0 3.0
3.0 0.0,
2x2 PairwiseListMatrices.PairwiseListMatrix{Float64,false}:
0.0 1.0
1.0 0.0)
julia> join(l, r, kind=:outer) # a b c d
(
4x4 PairwiseListMatrices.PairwiseListMatrix{Float64,false}:
0.0 1.0 2.0 NaN
1.0 0.0 3.0 NaN
2.0 3.0 0.0 NaN
NaN NaN NaN NaN,
4x4 PairwiseListMatrices.PairwiseListMatrix{Float64,false}:
NaN NaN NaN NaN
NaN 0.0 1.0 2.0
NaN 1.0 0.0 3.0
NaN 2.0 3.0 0.0)
```
"""
function join{L <: AbstractFloat, R <: AbstractFloat, DL, DR}(left::PairwiseListMatrix{L, DL}, right::PairwiseListMatrix{R, DR}; kind::Symbol = :inner ,
missing::Tuple{L,R} = (L(NaN), R(NaN)) )

labels_left = labels(left)
labels_right = labels(right)

if kind == :inner
out_labels = intersect(labels_left, labels_right)
N = length(out_labels)
out_L = PairwiseListMatrix(L, N, out_labels, DL)
out_R = PairwiseListMatrix(R, N, out_labels, DR)
for i in 1:N
li = out_labels[i]
for j in i:N
lj = out_labels[j]
out_L[i,j] = getlabel(left, li, lj)
out_R[i,j] = getlabel(right, li, lj)
end
end
return(out_L, out_R)
elseif kind == :left
out_labels = labels_left
N = length(out_labels)
out_R = PairwiseListMatrix(R, N, out_labels, DR)
for i in 1:N
li = out_labels[i]
flag_i_r = li in labels_right
for j in i:N
lj = out_labels[j]
out_R[i,j] = (flag_i_r && (lj in labels_right)) ? getlabel(right, li, lj) : missing[2]
end
end
return(left, out_R)
elseif kind == :right
out_labels = labels_right
N = length(out_labels)
out_L = PairwiseListMatrix(L, N, out_labels, DL)
for i in 1:N
li = out_labels[i]
flag_i_l = li in labels_left
for j in i:N
lj = out_labels[j]
out_L[i,j] = (flag_i_l && (lj in labels_left)) ? getlabel(left, li, lj) : missing[1]
end
end
return(out_L, right)
elseif kind == :outer
out_labels = union(labels_left, labels_right)
N = length(out_labels)
out_L = PairwiseListMatrix(L, N, out_labels, DL)
out_R = PairwiseListMatrix(R, N, out_labels, DR)
for i in 1:N
li = out_labels[i]
flag_i_l = li in labels_left
flag_i_r = li in labels_right
for j in i:N
lj = out_labels[j]
out_L[i,j] = (flag_i_l && (lj in labels_left)) ? getlabel(left, li, lj) : missing[1]
out_R[i,j] = (flag_i_r && (lj in labels_right)) ? getlabel(right, li, lj) : missing[2]
end
end
return(out_L, out_R)
else
throw(ArgumentError("Unknown kind of join requested: use :inner, :left, :right or :outer"))
end
end
38 changes: 38 additions & 0 deletions test/runtests.jl
View file
Original file line number Diff line number Diff line change
Expand Up @@ -543,6 +543,44 @@ let values = [1,2,3,4,5,6],

end

print("""
join
----
""")

let left = PairwiseListMatrix(collect(1.:15.), Char['a','b','c','d','e'], true),
right = PairwiseListMatrix(collect(1.:10.), Char['a','e','i','o','u'], false)

a, b = join(left, right)
@test (a, b) == join(left, right, kind=:inner)
@test labels(a) == labels(b)
@test labels(a) == ['a', 'e']
@test size(a) == (2,2)
@test size(b) == (2,2)

a, b = join(left, right, kind=:left)
@test labels(a) == labels(b)
@test labels(a) == labels(left)
@test a == left
@test b != right

a, b = join(left, right, kind=:right)
@test labels(a) == labels(b)
@test labels(b) == labels(right)
@test a != left
@test b == right

a, b = join(left, right, kind=:outer)
@test labels(a) == labels(b)
@test labels(a) == ['a', 'b', 'c', 'd', 'e', 'i', 'o', 'u']
@test size(a) == (8,8)
@test size(b) == (8,8)
@test a != left
@test b != right

end

# #### DataFrames #### #

if dataframes_installed
Expand Down

0 comments on commit 0a85c61

Please sign in to comment.