Performance tips#

using DataFrames
using BenchmarkTools
using CategoricalArrays
using PooledArrays
using Random

Access by column number is faster than by name#

x = DataFrame(rand(5, 1000), :auto)
@btime $x[!, 500]; ## Faster
  2.785 ns (0 allocations: 0 bytes)
@btime $x.x500;  ## Slower
  11.834 ns (0 allocations: 0 bytes)

When working with data DataFrame use barrier functions or type annotation#

function f_bad() ## this function will be slow
    Random.seed!(1)
    x = DataFrame(rand(1000000, 2), :auto)
    y, z = x[!, 1], x[!, 2]
    p = 0.0
    for i in 1:nrow(x)
        p += y[i] * z[i]
    end
    p
end

@btime f_bad();
# if you run @code_warntype f_bad() then you notice
# that Julia does not know column types of `DataFrame`
  97.949 ms (5999022 allocations: 122.06 MiB)

solution 1 is to use barrier function (it should be possible to use it in almost any code) for the calculation. You will notice much less memopry allocations and faster performance.

function f_inner(y, z)
    p = 0.0
    for i in eachindex(y, z)
        p += y[i] * z[i]
    end
    p
end

function f_barrier()
    Random.seed!(1)
    x = DataFrame(rand(1000000, 2), :auto)
    f_inner(x[!, 1], x[!, 2])
end

@btime f_barrier();
  3.903 ms (44 allocations: 30.52 MiB)

or use inbuilt function if possible

using LinearAlgebra

function f_inbuilt()
    Random.seed!(1)
    x = DataFrame(rand(1000000, 2), :auto)
    dot(x[!, 1], x[!, 2])
end

@btime f_inbuilt();
  3.322 ms (44 allocations: 30.52 MiB)

solution 2 is to provide the types of extracted columns. However, there are cases in which you will not know these types.

function f_typed()
    Random.seed!(1)
    x = DataFrame(rand(1000000, 2), :auto)
    y::Vector{Float64}, z::Vector{Float64} = x[!, 1], x[!, 2]
    p = 0.0
    for i in 1:nrow(x)
        p += y[i] * z[i]
    end
    p
end

@btime f_typed();
  3.870 ms (44 allocations: 30.52 MiB)

In general for tall and narrow tables it is often useful to use Tables.rowtable, Tables.columntable or Tables.namedtupleiterator for intermediate processing of data in a type-stable way.

Consider using delayed DataFrame creation technique#

also notice the difference in performance between copying vs non-copying data frame creation

function f1()
    x = DataFrame([Vector{Float64}(undef, 10^4) for i in 1:100], :auto, copycols=false) ## we work with a DataFrame directly
    for c in 1:ncol(x)
        d = x[!, c]
        for r in 1:nrow(x)
            d[r] = rand()
        end
    end
    x
end

function f1a()
    x = DataFrame([Vector{Float64}(undef, 10^4) for i in 1:100], :auto) ## we work with a DataFrame directly
    for c in 1:ncol(x)
        d = x[!, c]
        for r in 1:nrow(x)
            d[r] = rand()
        end
    end
    x
end

function f2()
    x = Vector{Any}(undef, 100)
    for c in 1:length(x)
        d = Vector{Float64}(undef, 10^4)
        for r in eachindex(d)
            d[r] = rand()
        end
        x[c] = d
    end
    DataFrame(x, :auto, copycols=false) ## we delay creation of DataFrame after we have our job done
end

function f2a()
    x = Vector{Any}(undef, 100)
    for c in eachindex(x)
        d = Vector{Float64}(undef, 10^4)
        for r in eachindex(d)
            d[r] = rand()
        end
        x[c] = d
    end
    DataFrame(x, :auto) ## we delay creation of DataFrame after we have our job done
end

@btime f1();
@btime f1a();
@btime f2();
@btime f2a();
  26.094 ms (1949728 allocations: 37.40 MiB)
  27.518 ms (1950028 allocations: 45.03 MiB)
  1.207 ms (728 allocations: 7.66 MiB)
  1.648 ms (1028 allocations: 15.29 MiB)

You can add rows to a DataFrame in place and it is fast#

x = DataFrame(rand(10^6, 5), :auto)
y = DataFrame(transpose(1.0:5.0), :auto)
z = [1.0:5.0;]

@btime vcat($x, $y); ## creates a new DataFrame - slow
@btime append!($x, $y); ## in place - fast

x = DataFrame(rand(10^6, 5), :auto) ## reset to the same starting point
@btime push!($x, $z); ## add a single row in place - fast
  2.735 ms (213 allocations: 38.16 MiB)
  1.099 μs (30 allocations: 1.52 KiB)
  429.291 ns (16 allocations: 256 bytes)

Allowing missing as well as categorical slows down computations#

using StatsBase

function test(data) ## uses countmap function to test performance
    println(eltype(data))
    x = rand(data, 10^6)
    y = categorical(x)
    println(" raw:")
    @btime countmap($x)
    println(" categorical:")
    @btime countmap($y)
    nothing
end

test(1:10)
test([randstring() for i in 1:10])
test(allowmissing(1:10))
test(allowmissing([randstring() for i in 1:10]))
Int64
 raw:
  1.932 ms (8 allocations: 7.63 MiB)
 categorical:
  16.107 ms (1000004 allocations: 30.52 MiB)
String
 raw:
  23.543 ms (4 allocations: 448 bytes)
 categorical:
  33.864 ms (1000004 allocations: 30.52 MiB)
Union{Missing, Int64}
 raw:
  7.415 ms (4 allocations: 464 bytes)
 categorical:
  16.489 ms (1000004 allocations: 30.52 MiB)
Union{Missing, String}
 raw:
  17.170 ms (4 allocations: 448 bytes)
 categorical:
  27.073 ms (1000004 allocations: 30.52 MiB)

When aggregating use column selector and prefer integer, categorical, or pooled array grouping variable#

df = DataFrame(x=rand('a':'d', 10^7), y=1);
gdf = groupby(df, :x)

GroupedDataFrame with 4 groups based on key: x

First Group (2499624 rows): x = 'a': ASCII/Unicode U+0061 (category Ll: Letter, lowercase)
2499599 rows omitted
Rowxy
CharInt64
1a1
2a1
3a1
4a1
5a1
6a1
7a1
8a1
9a1
10a1
11a1
12a1
13a1
2499613a1
2499614a1
2499615a1
2499616a1
2499617a1
2499618a1
2499619a1
2499620a1
2499621a1
2499622a1
2499623a1
2499624a1

Last Group (2499594 rows): x = 'b': ASCII/Unicode U+0062 (category Ll: Letter, lowercase)
2499569 rows omitted
Rowxy
CharInt64
1b1
2b1
3b1
4b1
5b1
6b1
7b1
8b1
9b1
10b1
11b1
12b1
13b1
2499583b1
2499584b1
2499585b1
2499586b1
2499587b1
2499588b1
2499589b1
2499590b1
2499591b1
2499592b1
2499593b1
2499594b1

traditional syntax, slow

@btime combine(v -> sum(v.y), $gdf)
  17.181 ms (333 allocations: 19.08 MiB)
4×2 DataFrame
Rowxx1
CharInt64
1a2499624
2c2499816
3d2500966
4b2499594

use column selector

@btime combine($gdf, :y => sum)
  7.076 ms (199 allocations: 9.41 KiB)
4×2 DataFrame
Rowxy_sum
CharInt64
1a2499624
2c2499816
3d2500966
4b2499594
transform!(df, :x => categorical => :x);
gdf = groupby(df, :x)

GroupedDataFrame with 4 groups based on key: x

First Group (2499624 rows): x = CategoricalArrays.CategoricalValue{Char, UInt32} 'a'
2499599 rows omitted
Rowxy
Cat…Int64
1a1
2a1
3a1
4a1
5a1
6a1
7a1
8a1
9a1
10a1
11a1
12a1
13a1
2499613a1
2499614a1
2499615a1
2499616a1
2499617a1
2499618a1
2499619a1
2499620a1
2499621a1
2499622a1
2499623a1
2499624a1

Last Group (2500966 rows): x = CategoricalArrays.CategoricalValue{Char, UInt32} 'd'
2500941 rows omitted
Rowxy
Cat…Int64
1d1
2d1
3d1
4d1
5d1
6d1
7d1
8d1
9d1
10d1
11d1
12d1
13d1
2500955d1
2500956d1
2500957d1
2500958d1
2500959d1
2500960d1
2500961d1
2500962d1
2500963d1
2500964d1
2500965d1
2500966d1
@btime combine($gdf, :y => sum)
  7.012 ms (207 allocations: 9.89 KiB)
4×2 DataFrame
Rowxy_sum
Cat…Int64
1a2499624
2b2499594
3c2499816
4d2500966
transform!(df, :x => PooledArray{Char} => :x)
10000000×2 DataFrame
9999975 rows omitted
Rowxy
CharInt64
1a1
2c1
3c1
4c1
5d1
6a1
7c1
8b1
9b1
10b1
11a1
12b1
13b1
9999989b1
9999990b1
9999991b1
9999992c1
9999993b1
9999994a1
9999995d1
9999996c1
9999997d1
9999998b1
9999999c1
10000000d1
gdf = groupby(df, :x)

GroupedDataFrame with 4 groups based on key: x

First Group (2499624 rows): x = 'a': ASCII/Unicode U+0061 (category Ll: Letter, lowercase)
2499599 rows omitted
Rowxy
CharInt64
1a1
2a1
3a1
4a1
5a1
6a1
7a1
8a1
9a1
10a1
11a1
12a1
13a1
2499613a1
2499614a1
2499615a1
2499616a1
2499617a1
2499618a1
2499619a1
2499620a1
2499621a1
2499622a1
2499623a1
2499624a1

Last Group (2499594 rows): x = 'b': ASCII/Unicode U+0062 (category Ll: Letter, lowercase)
2499569 rows omitted
Rowxy
CharInt64
1b1
2b1
3b1
4b1
5b1
6b1
7b1
8b1
9b1
10b1
11b1
12b1
13b1
2499583b1
2499584b1
2499585b1
2499586b1
2499587b1
2499588b1
2499589b1
2499590b1
2499591b1
2499592b1
2499593b1
2499594b1
@btime combine($gdf, :y => sum)
  7.083 ms (201 allocations: 9.47 KiB)
4×2 DataFrame
Rowxy_sum
CharInt64
1a2499624
2c2499816
3d2500966
4b2499594

Use views instead of materializing a new DataFrame#

x = DataFrame(rand(100, 1000), :auto)
@btime $x[1:1, :]
  472.793 μs (3015 allocations: 143.79 KiB)
1×1000 DataFrame
900 columns omitted
Rowx1x2x3x4x5x6x7x8x9x10x11x12x13x14x15x16x17x18x19x20x21x22x23x24x25x26x27x28x29x30x31x32x33x34x35x36x37x38x39x40x41x42x43x44x45x46x47x48x49x50x51x52x53x54x55x56x57x58x59x60x61x62x63x64x65x66x67x68x69x70x71x72x73x74x75x76x77x78x79x80x81x82x83x84x85x86x87x88x89x90x91x92x93x94x95x96x97x98x99x100
Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64
10.1706760.4825240.1653710.286710.9758930.1951690.8153530.7995850.2535670.5333030.5255010.2302380.5315160.009575150.7459030.7928150.7283190.1334420.9814230.469010.6412190.7875870.6927590.5956660.4960430.174510.6428360.2150830.1560660.9376620.7883490.4467310.02647510.6081870.04960190.2091620.3184990.2860480.3820110.2632190.2431710.7720680.8912770.1266650.174460.1521570.4281550.1201950.5340410.774090.6977850.1513860.4567280.2531050.04949990.8055950.5530760.687420.7322190.514020.1488610.4048230.3834180.5128490.0833240.417380.9036770.06805980.4370730.8835550.6337130.09277440.7462470.7181360.2205640.1900880.6735320.8450.8641760.09402940.254630.7296720.9111440.2071480.8971710.500630.6437390.5243160.3052170.971150.488950.05118320.6415210.008188340.05421990.1464980.9418230.2163590.2624210.769855
@btime $x[1, :]
  22.904 ns (0 allocations: 0 bytes)
DataFrameRow (1000 columns)
900 columns omitted
Rowx1x2x3x4x5x6x7x8x9x10x11x12x13x14x15x16x17x18x19x20x21x22x23x24x25x26x27x28x29x30x31x32x33x34x35x36x37x38x39x40x41x42x43x44x45x46x47x48x49x50x51x52x53x54x55x56x57x58x59x60x61x62x63x64x65x66x67x68x69x70x71x72x73x74x75x76x77x78x79x80x81x82x83x84x85x86x87x88x89x90x91x92x93x94x95x96x97x98x99x100
Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64
10.1706760.4825240.1653710.286710.9758930.1951690.8153530.7995850.2535670.5333030.5255010.2302380.5315160.009575150.7459030.7928150.7283190.1334420.9814230.469010.6412190.7875870.6927590.5956660.4960430.174510.6428360.2150830.1560660.9376620.7883490.4467310.02647510.6081870.04960190.2091620.3184990.2860480.3820110.2632190.2431710.7720680.8912770.1266650.174460.1521570.4281550.1201950.5340410.774090.6977850.1513860.4567280.2531050.04949990.8055950.5530760.687420.7322190.514020.1488610.4048230.3834180.5128490.0833240.417380.9036770.06805980.4370730.8835550.6337130.09277440.7462470.7181360.2205640.1900880.6735320.8450.8641760.09402940.254630.7296720.9111440.2071480.8971710.500630.6437390.5243160.3052170.971150.488950.05118320.6415210.008188340.05421990.1464980.9418230.2163590.2624210.769855
@btime view($x, 1:1, :)
  22.843 ns (0 allocations: 0 bytes)
1×1000 SubDataFrame
900 columns omitted
Rowx1x2x3x4x5x6x7x8x9x10x11x12x13x14x15x16x17x18x19x20x21x22x23x24x25x26x27x28x29x30x31x32x33x34x35x36x37x38x39x40x41x42x43x44x45x46x47x48x49x50x51x52x53x54x55x56x57x58x59x60x61x62x63x64x65x66x67x68x69x70x71x72x73x74x75x76x77x78x79x80x81x82x83x84x85x86x87x88x89x90x91x92x93x94x95x96x97x98x99x100
Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64
10.1706760.4825240.1653710.286710.9758930.1951690.8153530.7995850.2535670.5333030.5255010.2302380.5315160.009575150.7459030.7928150.7283190.1334420.9814230.469010.6412190.7875870.6927590.5956660.4960430.174510.6428360.2150830.1560660.9376620.7883490.4467310.02647510.6081870.04960190.2091620.3184990.2860480.3820110.2632190.2431710.7720680.8912770.1266650.174460.1521570.4281550.1201950.5340410.774090.6977850.1513860.4567280.2531050.04949990.8055950.5530760.687420.7322190.514020.1488610.4048230.3834180.5128490.0833240.417380.9036770.06805980.4370730.8835550.6337130.09277440.7462470.7181360.2205640.1900880.6735320.8450.8641760.09402940.254630.7296720.9111440.2071480.8971710.500630.6437390.5243160.3052170.971150.488950.05118320.6415210.008188340.05421990.1464980.9418230.2163590.2624210.769855
@btime $x[1:1, 1:20]
  9.859 μs (70 allocations: 3.09 KiB)
1×20 DataFrame
Rowx1x2x3x4x5x6x7x8x9x10x11x12x13x14x15x16x17x18x19x20
Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64
10.1706760.4825240.1653710.286710.9758930.1951690.8153530.7995850.2535670.5333030.5255010.2302380.5315160.009575150.7459030.7928150.7283190.1334420.9814230.46901
@btime $x[1, 1:20]
  25.006 ns (0 allocations: 0 bytes)
DataFrameRow (20 columns)
Rowx1x2x3x4x5x6x7x8x9x10x11x12x13x14x15x16x17x18x19x20
Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64
10.1706760.4825240.1653710.286710.9758930.1951690.8153530.7995850.2535670.5333030.5255010.2302380.5315160.009575150.7459030.7928150.7283190.1334420.9814230.46901
@btime view($x, 1:1, 1:20)
  24.956 ns (0 allocations: 0 bytes)
1×20 SubDataFrame
Rowx1x2x3x4x5x6x7x8x9x10x11x12x13x14x15x16x17x18x19x20
Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64Float64
10.1706760.4825240.1653710.286710.9758930.1951690.8153530.7995850.2535670.5333030.5255010.2302380.5315160.009575150.7459030.7928150.7283190.1334420.9814230.46901

This notebook was generated using Literate.jl.