Possible pitfalls

using DataFrames

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	0.587516	0.718981	0.62205	0.791086	0.505925
2	0.900673	0.0775141	0.210494	0.15541	0.0177345
3	0.436133	0.866412	0.603696	0.360206	0.424278

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 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 (wide tables),

• but Tables.namedtupleiterator for fast execution (tall tables)

this table is wide

df1 = DataFrame([rand([1:2, 'a':'b', false:true, 1.0:2.0]) for i in 1:900], :auto)

2×900 DataFrame

800 columns omitted

Row	x1	x2	x3	x4	x5	x6	x7	x8	x9	x10	x11	x12	x13	x14	x15	x16	x17	x18	x19	x20	x21	x22	x23	x24	x25	x26	x27	x28	x29	x30	x31	x32	x33	x34	x35	x36	x37	x38	x39	x40	x41	x42	x43	x44	x45	x46	x47	x48	x49	x50	x51	x52	x53	x54	x55	x56	x57	x58	x59	x60	x61	x62	x63	x64	x65	x66	x67	x68	x69	x70	x71	x72	x73	x74	x75	x76	x77	x78	x79	x80	x81	x82	x83	x84	x85	x86	x87	x88	x89	x90	x91	x92	x93	x94	x95	x96	x97	x98	x99	x100	⋯
	Bool	Char	Bool	Int64	Int64	Int64	Float64	Char	Int64	Int64	Char	Float64	Float64	Int64	Char	Int64	Char	Int64	Int64	Float64	Char	Float64	Bool	Char	Int64	Bool	Int64	Float64	Float64	Char	Float64	Bool	Int64	Bool	Char	Int64	Int64	Int64	Char	Int64	Float64	Int64	Bool	Int64	Char	Float64	Bool	Bool	Bool	Float64	Char	Float64	Char	Char	Char	Bool	Float64	Char	Char	Char	Float64	Char	Int64	Bool	Bool	Char	Bool	Float64	Float64	Int64	Int64	Char	Float64	Char	Bool	Int64	Float64	Float64	Float64	Int64	Int64	Char	Int64	Bool	Int64	Char	Char	Float64	Int64	Bool	Char	Float64	Int64	Int64	Char	Bool	Float64	Float64	Int64	Float64	⋯
1	false	a	false	1	1	1	1.0	a	1	1	a	1.0	1.0	1	a	1	a	1	1	1.0	a	1.0	false	a	1	false	1	1.0	1.0	a	1.0	false	1	false	a	1	1	1	a	1	1.0	1	false	1	a	1.0	false	false	false	1.0	a	1.0	a	a	a	false	1.0	a	a	a	1.0	a	1	false	false	a	false	1.0	1.0	1	1	a	1.0	a	false	1	1.0	1.0	1.0	1	1	a	1	false	1	a	a	1.0	1	false	a	1.0	1	1	a	false	1.0	1.0	1	1.0	⋯
2	true	b	true	2	2	2	2.0	b	2	2	b	2.0	2.0	2	b	2	b	2	2	2.0	b	2.0	true	b	2	true	2	2.0	2.0	b	2.0	true	2	true	b	2	2	2	b	2	2.0	2	true	2	b	2.0	true	true	true	2.0	b	2.0	b	b	b	true	2.0	b	b	b	2.0	b	2	true	true	b	true	2.0	2.0	2	2	b	2.0	b	true	2	2.0	2.0	2.0	2	2	b	2	true	2	b	b	2.0	2	true	b	2.0	2	2	b	true	2.0	2.0	2	2.0	⋯

@time collect(eachrow(df1));

  0.041341 seconds (52.71 k allocations: 3.614 MiB, 99.91% compilation time)

@time collect(Tables.namedtupleiterator(df1));

  7.539398 seconds (885.91 k allocations: 70.446 MiB, 99.68% compilation time)

as you can see the time to compile Tables.namedtupleiterator is very large in this case, and it would get much worse if the table was wider (that is why we include this tip in pitfalls notebook)

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	0.791452	0.297286	0.049207	0.88973	0.309781	0.124727	0.526994	0.376758	0.982059	0.201495
2	0.220364	0.701363	0.37604	0.853463	0.292551	0.160833	0.403648	0.806145	0.686896	0.319294
3	0.375985	0.0946538	0.401264	0.259818	0.426253	0.992351	0.351296	0.897025	0.642545	0.36698
4	0.190554	0.799914	0.814485	0.664285	0.243209	0.538027	0.281022	0.716538	0.735261	0.0711319
5	0.794052	0.707524	0.982855	0.679856	0.33846	0.377405	0.393751	0.78178	0.44972	0.205827
6	0.388176	0.641699	0.793341	0.455715	0.24483	0.519513	0.697198	0.371501	0.707692	0.252749
7	0.432196	0.0478298	0.691303	0.952995	0.352697	0.359571	0.597546	0.151098	0.413731	0.00542862
8	0.796189	0.0284523	0.102914	0.570933	0.207063	0.212477	0.598974	0.391049	0.290673	0.416863
9	0.514341	0.574274	0.00955344	0.0426117	0.874077	0.198039	0.504121	0.848645	0.607743	0.464585
10	0.678113	0.601279	0.30789	0.037617	0.198317	0.0906097	0.0780964	0.339857	0.0657225	0.0773766
11	0.216809	0.91802	0.663682	0.813137	0.988659	0.0123706	0.322914	0.79604	0.382519	0.227306
12	0.952948	0.797773	0.919703	0.983742	0.976059	0.818607	0.29084	0.657554	0.921209	0.362167
13	0.370819	0.777084	0.343126	0.457476	0.411953	0.506055	0.829612	0.894446	0.113823	0.804729
⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮
999989	0.59812	0.341671	0.756817	0.129677	0.558078	0.119235	0.662959	0.619881	0.902543	0.332156
999990	0.795765	0.528571	0.711405	0.0577124	0.930984	0.551859	0.703604	0.269038	0.557876	0.433196
999991	0.579624	0.233688	0.215393	0.304601	0.851172	0.204315	0.0941204	0.294733	0.745845	0.504573
999992	0.495098	0.165949	0.462764	0.83711	0.11577	0.769074	0.10481	0.630765	0.154507	0.641631
999993	0.695556	0.318069	0.398515	0.740937	0.858254	0.235751	0.184672	0.0391809	0.419516	0.183319
999994	0.165754	0.677366	0.134086	0.644506	0.594977	0.0533597	0.616334	0.391897	0.161337	0.967639
999995	0.168486	0.869069	0.0727793	0.0418242	0.226054	0.0179684	0.614555	0.911016	0.256836	0.658037
999996	0.0873745	0.77549	0.224858	0.176993	0.387384	0.100599	0.333415	0.283454	0.292832	0.447891
999997	0.792855	0.471462	0.23574	0.18291	0.235345	0.801913	0.563771	0.558161	0.508129	0.595255
999998	0.82707	0.813067	0.0496293	0.81687	0.0583434	0.825218	0.63372	0.902962	0.823672	0.417822
999999	0.805874	0.665029	0.376935	0.849447	0.672471	0.246654	0.568449	0.506283	0.0738984	0.650809
1000000	0.357941	0.279164	0.286021	0.229009	0.569281	0.348493	0.592393	0.922544	0.825604	0.48944

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

  2.127519 seconds (60.08 M allocations: 1.056 GiB, 6.17% gc time, 3.69% compilation time)

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

  2.119773 seconds (59.99 M allocations: 1.050 GiB, 6.50% gc time)

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

  0.195608 seconds (200.06 k allocations: 21.156 MiB, 24.98% gc time, 96.98% compilation time)

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

  0.005788 seconds (22 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

  0.392389 seconds (509.34 k allocations: 41.875 MiB, 98.52% compilation time: 93% of which was recompilation)

1000000-element Vector{Float64}:
 4.549489608990553
 4.820598260856793
 4.808170226496026
 5.054427780801031
 5.711228368654866
 5.0724123056568216
 4.00439573440345
 3.61558727488398
 4.637989455369663
 2.474878422732406
 ⋮
 4.377478339806004
 4.073770679842652
 4.40725544523935
 3.836624901881449
 3.110289532802179
 4.945542466292135
 6.168372468024979
 5.415848010904047
 4.899889662604407

Do it again

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

  0.004722 seconds (125 allocations: 7.635 MiB)

1000000-element Vector{Float64}:
 4.549489608990553
 4.820598260856793
 4.808170226496026
 5.054427780801031
 5.711228368654866
 5.0724123056568216
 4.00439573440345
 3.61558727488398
 4.637989455369663
 2.474878422732406
 ⋮
 4.377478339806004
 4.073770679842652
 4.40725544523935
 3.836624901881449
 3.110289532802179
 4.945542466292135
 6.168372468024979
 5.415848010904047
 4.899889662604407

---

This notebook was generated using Literate.jl.