When all you want is to just tally.
Status
Table of contents
Installation
Since Tally.jl is a registered package, it can be simply installed as follows:
julia> using Pkg; Pkg.install("Tally")Usage
Creating a tally (frequency count)
Given some data stored in an object data, one can count the number of occurrences of items by calling tally(data):
julia> T = tally(["x", "x", "y", "x"])
Tally with 4 items in 2 groups:
x | 3 | 75%
y | 1 | 25%One can put in any iterable object
julia> T = tally(rand(-1:1, 10, 10)) # a random 10x10 matrix with entries in [-1, 0, 1]
Tally with 100 items in 3 groups:
-1 | 37 | 37%
1 | 32 | 32%
0 | 31 | 31%A tally can be extended by adding more items via push! or append!.
julia> push!(T, "x")
Tally with 5 items in 2 groups:
x | 4 | 80%
y | 1 | 20%
julia> append!(T, ["x", "y", "y"])
Tally with 8 items in 2 groups:
x | 5 | 62%
y | 3 | 38%Plotting a tally
Tally.jl comes with some basic plotting functionalities to plot tallies within the julia REPL:
julia> T = tally(rand(-1:1, 10, 10))
Tally with 100 items in 3 groups:
0 | 43 | 43%
1 | 33 | 33%
-1 | 24 | 24%
julia> Tally.plot(T)
┌ ┐
0 ┤■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■ 43 43%
1 ┤■■■■■■■■■■■■■■■■■■■■■■■■■■■■ 33 33%
-1 ┤■■■■■■■■■■■■■■■■■■■■ 24 24%
└ ┘See ?Tally.plot for a list of options on how to customize this plot, which includes giving it a title or choosing a different ordering.
Prison count
For tallies with counts not too large and pure entertainment, one can also plot tallies using a "prison count":
julia> T = tally([1, 1, 1, 2, 2, 2, 2, 2, 2, -1]);
julia> prison_count(T)
2 ┃ ┼┼┼┼ │
━━━╋━━━━━━━
1 ┃ │││
━━━╋━━━━━━━
-1 ┃ │Tables.jl interface and printing using PrettyTable.jl
The objects constructed by Tally.jl implement the Tables.jl interface and thus can be printed using PrettyTable.jl:
julia> T = tally([1, 1, 1, 2, 2, 2, 2, 2, 2, -1]);
julia> pretty_table(T)
┌───────┬────────┐
│ Keys │ Values │
│ Int64 │ Int64 │
├───────┼────────┤
│ 2 │ 6 │
│ 1 │ 3 │
│ -1 │ 1 │
└───────┴────────┘Plotting a tally using Plots.jl
If you have Plots.jl installed and loaded, you can also plot it using this functionality:
julia> T = tally(rand(-1:1, 10, 10))
Tally with 100 items in 3 groups:
1 | 38 | 38%
0 | 34 | 34%
-1 | 28 | 28%
julia> bar(T, legend = false)This will produce:
See ?Plots.bar for more information on how to customize this plot.
Advanced usage
Tally is too slow
To work also for objects for which a consistent hash is not implemented, tally does not use hash by default. This can be enabled using the use_hash = true keyword.
julia> v = [rand([[1], [2]]) for i in 1:100000];
julia> @btime tally($v)
14.563 ms (100005 allocations: 1.53 MiB)
Tally with 100000 items in 2 groups:
[2] | 50146 | 50.1%
[1] | 49854 | 49.9%
julia> @btime tally($v, use_hash = true)
2.183 ms (7 allocations: 720 bytes)
Tally with 100000 items in 2 groups:
[2] | 50146 | 50.1%
[1] | 49854 | 49.9%Counting up to an equivalence
When counting, sometimes one wants to do a tally only with respect to some other invariant or with respect to an equivalence relation different from ==. For this task tally provides the by and equivalence keyword arguments. The function tally will consider two elements x, y from the input collection equal when counting, whenever equivalence(by(x), by(y)) is true. The default values are by = identity and equivalence = isequal. If equivalence does not define an equivalence relation, the result will be nonsense.
Note that to indicate that the counting is non-standard, Tally will print the objects within square brackets [ ].
julia> v = 1:100000;
julia> tally(v, by = iseven)
Tally with 100000 items in 2 groups:
[2] | 50000 | 50%
[1] | 50000 | 50%
julia> tally(v, by = x -> mod(x, 3))
Tally with 100000 items in 3 groups:
[1] | 33334 | 33.334%
[3] | 33333 | 33.333%
[2] | 33333 | 33.333%
julia> v = ["abb", "ba", "aa", "ba", "bbba", "aaab"];
julia> tally(v, equivalence = (x, y) -> first(x) == first(y) && last(x) == last(y))
Tally with 6 items in 3 groups:
[ba] | 3 | 50%
[abb] | 2 | 33%
[aa] | 1 | 17%The optional equivalence argument is important in case the equivalence relation under consideration does not admit easily computable unique representatives. Here is a real world example using Hecke.jl, where we only want to count algebraic objects up to isomorphism and thus can make use of the equivalence functionality. We make a tally of the 2-parts of the class groups of the first imaginary quadratic number fields:
julia> using Hecke;
julia> ds = Hecke.squarefree_up_to(1000);
julia> T = tally((class_group(quadratic_field(-d)[1])[1] for d in ds), equivalence = (G, H) -> is_isomorphic(psylow_subgroup(G, 2)[1], psylow_subgroup(H, 2)[1])[1]);
julia> Tally.plot(T)
┌ ┐
[GrpAb: Z/2] ┤■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■ 148 24.3%
[GrpAb: (Z/2)^2] ┤■■■■■■■■■■■■■■■■■■■■■■■■■■■■■ 122 20.1%
[GrpAb: Z/1] ┤■■■■■■■■■■■■■■■■■■■■■ 89 14.6%
[GrpAb: Z/4] ┤■■■■■■■■■■■■■■■ 64 10.5%
[GrpAb: Z/2 x Z/4] ┤■■■■■■■■■■■ 48 7.9%
[GrpAb: Z/8] ┤■■■■■■■■■ 39 6.4%
[GrpAb: Z/2 x Z/8] ┤■■■■■■■■ 35 5.8%
[GrpAb: (Z/2)^3] ┤■■■■ 19 3.1%
[GrpAb: (Z/2)^2 x Z/4] ┤■■■■ 18 3.0%
[GrpAb: Z/16] ┤■■■ 13 2.1%
[GrpAb: Z/32] ┤■■ 7 1.2%
[GrpAb: Z/2 x Z/16] ┤■ 5 0.8%
[GrpAb: (Z/2)^2 x Z/8] ┤ 1 0.2%
└ ┘Lazy tallies and animations
For maximal showoff potential, one can also construct "lazy" tallies, which can be plotted as an animation.
julia>T = lazy_tally((rand(-1:1) for i in 1:100));
julia> Tally.animate(T, badges = 4, delay = 0.2, title = "Will 0 win?")will yield
Note that a lazy tally T can be converted to an ordinary tally object by invoking materialize(T).