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	0.904476	0.829423	0.0503674	0.712011	0.452017
2	0.531043	0.133402	0.0159836	0.130781	0.312669
3	0.774213	0.260022	0.779513	0.786003	0.105047

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	0.478301	0.12516	0.443086	0.511008	0.881368	0.723646	0.891943	0.715338	0.226988	0.05517
2	0.537297	0.884842	0.761933	0.837568	0.142577	0.792199	0.602414	0.899725	0.594977	0.0328992
3	0.545046	0.146989	0.767474	0.750266	0.353807	0.901302	0.493169	0.289729	0.037967	0.601937
4	0.468204	0.860019	0.480399	0.433961	0.36144	0.0987604	0.869873	0.690603	0.0129994	0.858868
5	0.945626	0.737168	0.897333	0.721963	0.747643	0.378898	0.393668	0.965249	0.832653	0.368617
6	0.795077	0.24254	0.461301	0.071843	0.526477	0.885815	0.392322	0.804884	0.967799	0.484455
7	0.144368	0.0746826	0.296573	0.0946843	0.233691	0.207105	0.0322509	0.446124	0.316105	0.225936
8	0.246877	0.162375	0.829648	0.669706	0.255161	0.292926	0.255723	0.00901615	0.504461	0.0955296
9	0.601375	0.396815	0.630975	0.178445	0.373065	0.0397525	0.690713	0.504059	0.531599	0.243714
10	0.698136	0.881048	0.116188	0.486561	0.302637	0.0438733	0.503967	0.702356	0.273515	0.910538
11	0.0118833	0.856812	0.534104	0.0955872	0.191018	0.305049	0.689865	0.567668	0.320684	0.557577
12	0.191861	0.84547	0.239379	0.760855	0.362501	0.87176	0.729357	0.652585	0.76212	0.226549
13	0.172659	0.490196	0.286927	0.303485	0.00823402	0.120736	0.403559	0.755553	0.149911	0.605745
⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮
999989	0.25687	0.686048	0.141895	0.30186	0.749336	0.038252	0.688074	0.662057	0.655862	0.650975
999990	0.0755492	0.791997	0.615303	0.234143	0.874981	0.257686	0.833352	0.78647	0.0109449	0.341014
999991	0.465504	0.0621855	0.841689	0.589676	0.745905	0.359124	0.724153	0.480526	0.469831	0.730895
999992	0.643036	0.93191	0.992777	0.983357	0.459779	0.938882	0.776353	0.424043	0.234642	0.0562271
999993	0.160002	0.147143	0.496255	0.983457	0.276256	0.0725854	0.196438	0.210039	0.241376	0.978434
999994	0.951502	0.393141	0.449806	0.231952	0.275161	0.531019	0.845476	0.400511	0.293222	0.234725
999995	0.863946	0.468246	0.803938	0.588295	0.448115	0.466923	0.523416	0.492861	0.716858	0.131195
999996	0.814875	0.276221	0.46419	0.315931	0.413915	0.229485	0.889376	0.814128	0.0811651	0.0324598
999997	0.2221	0.669916	0.861596	0.87189	0.852959	0.325319	0.028256	0.54034	0.495246	0.832195
999998	0.183501	0.694183	0.236999	0.417095	0.0832853	0.364791	0.26945	0.420274	0.543252	0.0790998
999999	0.348597	0.0798918	0.500807	0.0939304	0.719459	0.142878	0.302941	0.391026	0.176512	0.981048
1000000	0.171675	0.261939	0.724141	0.796064	0.357761	0.792198	0.29389	0.269918	0.647273	0.575534

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

  2.084860 seconds (60.13 M allocations: 1.057 GiB, 8.12% gc time, 5.40% compilation time)

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

  1.929150 seconds (59.99 M allocations: 1.050 GiB, 8.39% gc time)

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

  0.356282 seconds (1.29 M allocations: 70.545 MiB, 5.93% gc time, 97.51% compilation time)

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

  0.008565 seconds (26 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

  0.604399 seconds (999.89 k allocations: 57.228 MiB, 97.41% compilation time: 95% of which was recompilation)

1000000-element Vector{Float64}:
 5.0520095448766185
 6.086430758355689
 4.887685872073017
 5.135126449042814
 6.988818506958208
 5.632513763521528
 2.071519953916969
 3.321422331055035
 4.190512286064625
 4.918819423163244
 ⋮
 6.441006218261728
 3.761986370440841
 4.606515314644607
 5.503791814623299
 4.331745999643407
 5.69981683021577
 3.29192976063087
 3.7370911066160897
 4.890392635140402

Do it again

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

  0.014927 seconds (127 allocations: 7.634 MiB)

1000000-element Vector{Float64}:
 5.0520095448766185
 6.086430758355689
 4.887685872073017
 5.135126449042814
 6.988818506958208
 5.632513763521528
 2.071519953916969
 3.321422331055035
 4.190512286064625
 4.918819423163244
 ⋮
 6.441006218261728
 3.761986370440841
 4.606515314644607
 5.503791814623299
 4.331745999643407
 5.69981683021577
 3.29192976063087
 3.7370911066160897
 4.890392635140402

---

This notebook was generated using Literate.jl.