Possible pitfalls

using DataFrames
using BenchmarkTools

Know what is copied when creating a DataFrame

x = DataFrame(rand(3, 5), :auto)

3×5 DataFrame

Row	x1	x2	x3	x4	x5
	Float64	Float64	Float64	Float64	Float64
1	0.2113	0.308308	0.0566407	0.828248	0.0765941
2	0.794645	0.998069	0.770412	0.653407	0.168957
3	0.139427	0.348479	0.123609	0.199557	0.735428

x and y are not the same object

y = copy(x)
x === y

false

x and y are not the same object

y = DataFrame(x)
x === y

false

the columns are also not the same

any(x[!, i] === y[!, i] for i in ncol(x))

false

x and y are not the same object

y = DataFrame(x, copycols=false)
x === y

false

But the columns are the same

all(x[!, i] === y[!, i] for i in ncol(x))

true

the same when creating data frames using kwarg syntax

x = 1:3;
y = [1, 2, 3];
df = DataFrame(x=x, y=y)

3×2 DataFrame

Row	x	y
	Int64	Int64
1	1	1
2	2	2
3	3	3

different object

y === df.y

false

range is converted to a vector

typeof(x), typeof(df.x)

(UnitRange{Int64}, Vector{Int64})

slicing rows always creates a copy

y === df[:, :y]

false

you can avoid copying by using copycols=false keyword argument in functions.

df = DataFrame(x=x, y=y, copycols=false)

3×2 DataFrame

Row	x	y
	Int64	Int64
1	1	1
2	2	2
3	3	3

now it is the same

y === df.y

true

not the same object

select(df, :y)[!, 1] === y

false

the same object

select(df, :y, copycols=false)[!, 1] === y

true

Do not modify the parent of GroupedDataFrame or view

x = DataFrame(id=repeat([1, 2], outer=3), x=1:6)
g = groupby(x, :id)

x[1:3, 1] = [2, 2, 2]
g ## well - it is wrong now, g is only a view

GroupedDataFrame with 2 groups based on key: id

First Group (3 rows): id = 2

Row	id	x
	Int64	Int64
1	2	1
2	2	3
3	1	5

⋮

Last Group (3 rows): id = 2

Row	id	x
	Int64	Int64
1	2	2
2	2	4
3	2	6

s = view(x, 5:6, :)

2×2 SubDataFrame

Row	id	x
	Int64	Int64
1	1	5
2	2	6

delete!(x, 3:6)

2×2 DataFrame

Row	id	x
	Int64	Int64
1	2	1
2	2	2

This is an error

s ## Will return BoundsError

Single column selection for a DataFrame

Single column selection for a DataFrame creates aliases with ! and getproperty syntax and copies with :

x = DataFrame(a=1:3)
x.b = x[!, 1] ## alias
x.c = x[:, 1] ## copy
x.d = x[!, 1][:] ## copy
x.e = copy(x[!, 1]) ## explicit copy
display(x)

3×5 DataFrame

Row	a	b	c	d	e
	Int64	Int64	Int64	Int64	Int64
1	1	1	1	1	1
2	2	2	2	2	2
3	3	3	3	3	3

x[1, 1] = 100
display(x)

3×5 DataFrame

Row	a	b	c	d	e
	Int64	Int64	Int64	Int64	Int64
1	100	100	1	1	1
2	2	2	2	2	2
3	3	3	3	3	3

When iterating rows of a data frame

• use eachrow to avoid compilation cost in wide tables,

• but Tables.namedtupleiterator for fast execution in tall tables

The table below is tall:

df2 = DataFrame(rand(10^6, 10), :auto)

1000000×10 DataFrame

999975 rows omitted

Row	x1	x2	x3	x4	x5	x6	x7	x8	x9	x10
	Float64	Float64	Float64	Float64	Float64	Float64	Float64	Float64	Float64	Float64
1	0.126173	0.91391	0.230886	0.46031	0.0800606	0.0696025	0.0561202	0.893639	0.894401	0.718733
2	0.660918	0.978677	0.611323	0.21381	0.328033	0.956103	0.587796	0.127799	0.406534	0.986028
3	0.75506	0.710277	0.653572	0.946084	0.628871	0.727277	0.635549	0.56663	0.397588	0.641642
4	0.969984	0.67806	0.992892	0.894194	0.393372	0.0337288	0.150722	0.0681556	0.379299	0.678804
5	0.266534	0.83572	0.304604	0.000771709	0.649048	0.279221	0.204479	0.598983	0.674805	0.521239
6	0.859502	0.371216	0.626017	0.658222	0.534773	0.481119	0.0269219	0.677936	0.116461	0.00703477
7	0.956322	0.439785	0.328789	0.618854	0.104963	0.216793	0.536987	0.248422	0.898859	0.0577803
8	0.285074	0.965578	0.34113	0.146301	0.279613	0.441333	0.0284473	0.658741	0.115377	0.482112
9	0.0183711	0.122434	0.374319	0.341599	0.107395	0.980639	0.736708	0.806391	0.748125	0.71007
10	0.380976	0.60499	0.83708	0.285092	0.420096	0.666504	0.24988	0.211287	0.245034	0.654536
11	0.62405	0.652209	0.760929	0.606618	0.111383	0.10651	0.0904231	0.666953	0.858432	0.778396
12	0.389306	0.950827	0.869131	0.204924	0.830092	0.716451	0.299398	0.45055	0.811171	0.486584
13	0.303644	0.411425	0.945022	0.838854	0.485448	0.0375779	0.234965	0.767782	0.835408	0.882997
⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮
999989	0.573598	0.60517	0.208363	0.113562	0.375677	0.407589	0.097919	0.510495	0.965661	0.687687
999990	0.0825845	0.274469	0.0438893	0.8678	0.100887	0.715827	0.729274	0.631956	0.492734	0.269479
999991	0.686946	0.593147	0.872832	0.353351	0.337731	0.23053	0.16554	0.364014	0.0924762	0.380706
999992	0.600088	0.0194345	0.361781	0.170843	0.373847	0.0926208	0.432099	0.629715	0.728548	0.753648
999993	0.84924	0.645248	0.0407621	0.404281	0.383247	0.228372	0.74692	0.535874	0.96791	0.59859
999994	0.874322	0.241312	0.723002	0.208205	0.545769	0.684953	0.905342	0.34449	0.699187	0.377778
999995	0.178979	0.243952	0.888587	0.963289	0.366592	0.68756	0.33175	0.103789	0.378261	0.350398
999996	0.845058	0.48191	0.408775	0.904383	0.846516	0.493856	0.0344712	0.581395	0.383492	0.159446
999997	0.710376	0.815444	0.21895	0.313038	0.668372	0.593656	0.494818	0.140116	0.668364	0.598582
999998	0.939286	0.564092	0.403675	0.924856	0.440322	0.011495	0.849363	0.93145	0.932504	0.249018
999999	0.582941	0.672818	0.848791	0.315112	0.636934	0.437168	0.324979	0.729984	0.202597	0.554824
1000000	0.11236	0.674445	0.411437	0.175847	0.249222	0.774572	0.901069	0.401976	0.122709	0.554017

@time map(sum, eachrow(df2));

  2.468679 seconds (60.12 M allocations: 1.056 GiB, 21.56% gc time, 4.90% compilation time)

@time map(sum, eachrow(df2));

  1.765101 seconds (59.99 M allocations: 1.050 GiB, 6.20% gc time)

@time map(sum, Tables.namedtupleiterator(df2));

  0.339066 seconds (1.22 M allocations: 66.508 MiB, 4.27% gc time, 94.07% compilation time)

@time map(sum, Tables.namedtupleiterator(df2));

  0.005196 seconds (21 allocations: 7.631 MiB)

as you can see - this time it is much faster to iterate a type stable container still you might want to use the select syntax, which is optimized for such reductions:

this includes compilation time

@time select(df2, AsTable(:) => ByRow(sum) => "sum").sum

  0.552321 seconds (711.57 k allocations: 43.126 MiB, 98.61% compilation time: 94% of which was recompilation)

1000000-element Vector{Float64}:
 4.443835677199177
 5.857019558194336
 6.662549653640113
 5.2392115580311875
 4.335405396831737
 4.359202661825034
 4.407556880885516
 3.7437040312173777
 4.946051256166412
 4.5554762944531015
 ⋮
 4.162624520064019
 5.400444138071531
 5.6043599690171435
 4.4931574135086505
 5.139301007663957
 5.221714755064596
 6.2460613897101105
 5.306147866526345
 4.377654217174633

Do it again

@time select(df2, AsTable(:) => ByRow(sum) => "sum").sum

  0.006508 seconds (123 allocations: 7.634 MiB)

1000000-element Vector{Float64}:
 4.443835677199177
 5.857019558194336
 6.662549653640113
 5.2392115580311875
 4.335405396831737
 4.359202661825034
 4.407556880885516
 3.7437040312173777
 4.946051256166412
 4.5554762944531015
 ⋮
 4.162624520064019
 5.400444138071531
 5.6043599690171435
 4.4931574135086505
 5.139301007663957
 5.221714755064596
 6.2460613897101105
 5.306147866526345
 4.377654217174633

---

This notebook was generated using Literate.jl.