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	0.529334	0.0597551	0.235795	0.501942	0.958109
2	0.484044	0.047142	0.114432	0.0418198	0.913857
3	0.500429	0.156764	0.0733438	0.843276	0.986249

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	0.812425	0.953875	0.289671	0.854766	0.509818	0.140464	0.08607	0.520695	0.205697	0.321225
2	0.208584	0.000966154	0.902822	0.423168	0.473373	0.157054	0.793769	0.957795	0.800943	0.664851
3	0.002694	0.326925	0.0368252	0.187928	0.126627	0.776543	0.402102	0.469324	0.0728587	0.844602
4	0.569028	0.31596	0.848604	0.812409	0.45489	0.86572	0.345673	0.236728	0.224588	0.646519
5	0.0631041	0.203245	0.701698	0.303583	0.674797	0.597116	0.606777	0.451958	0.39108	0.00379739
6	0.247021	0.809399	0.760111	0.598913	0.272918	0.597262	0.49311	0.520876	0.447351	0.292064
7	0.267973	0.211999	0.771751	0.639811	0.784383	0.121173	0.563557	0.549457	0.0545117	0.745306
8	0.108655	0.110102	0.953703	0.168199	0.976653	0.0276927	0.643222	0.963599	0.304311	0.678014
9	0.693502	0.54823	0.20262	0.210266	0.561789	0.900743	0.528027	0.615125	0.313564	0.557392
10	0.661105	0.661711	0.820527	0.440406	0.434706	0.429731	0.0608958	0.846089	0.422275	0.17791
11	0.234678	0.914611	0.715813	0.029249	0.257056	0.122383	0.262263	0.00461947	0.146143	0.0903481
12	0.193291	0.342759	0.914605	0.418826	0.135818	0.484762	0.191179	0.188863	0.364738	0.21938
13	0.242351	0.327346	0.448075	0.560953	0.592612	0.0503523	0.893915	0.835376	0.64798	0.246143
⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮
999989	0.661472	0.635782	0.792354	0.761077	0.0308713	0.099554	0.240259	0.35399	0.314355	0.686009
999990	0.0775762	0.845558	0.742922	0.464236	0.689336	0.548313	0.807805	0.671055	0.0915516	0.833302
999991	0.0976372	0.361425	0.714378	0.341477	0.0911392	0.403061	0.723864	0.777392	0.595327	0.25862
999992	0.501212	0.272561	0.0973808	0.331344	0.0614938	0.519877	0.839399	0.765083	0.688672	0.150719
999993	0.273739	0.401793	0.344902	0.693997	0.595919	0.83713	0.931485	0.0467357	0.24671	0.151542
999994	0.832169	0.267376	0.925687	0.418644	0.0413268	0.791828	0.732267	0.197472	0.778455	0.226904
999995	0.0660683	0.428659	0.340804	0.747841	0.279143	0.182118	0.698698	0.570114	0.918868	0.789216
999996	0.466766	0.64338	0.773518	0.738444	0.963	0.521375	0.100483	0.656539	0.82001	0.324571
999997	0.907574	0.59836	0.888056	0.378057	0.606351	0.0249071	0.0768992	0.531575	0.621098	0.440302
999998	0.612807	0.416059	0.728195	0.271598	0.571033	0.268268	0.230246	0.610069	0.474585	0.931925
999999	0.763149	0.923241	0.428408	0.99299	0.377282	0.43588	0.980952	0.00561584	0.65738	0.105073
1000000	0.645671	0.461747	0.28093	0.71416	0.38365	0.786039	0.162344	0.541735	0.13926	0.988997

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

  1.934170 seconds (60.11 M allocations: 1.056 GiB, 6.28% gc time, 6.29% compilation time)

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

  2.244732 seconds (59.99 M allocations: 1.050 GiB, 21.20% gc time)

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

  0.358052 seconds (1.32 M allocations: 73.135 MiB, 5.26% gc time, 98.21% compilation time)

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

  0.006158 seconds (25 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

  0.586411 seconds (662.47 k allocations: 41.440 MiB, 98.79% compilation time: 94% of which was recompilation)

1000000-element Vector{Float64}:
 4.694706502313453
 5.383324647550716
 3.2464293812878067
 5.320117297804017
 3.9971556695645343
 5.039026681750683
 4.709921099884204
 4.93415120764944
 5.131257550693444
 4.955355792546958
 ⋮
 4.227741340484659
 4.523953769215137
 5.2121280936199845
 5.021528576928276
 6.008086254445894
 5.073179235758298
 5.1147852166724865
 5.669970336464331
 5.1045325666422725

Do it again

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

  0.005626 seconds (127 allocations: 7.634 MiB)

1000000-element Vector{Float64}:
 4.694706502313453
 5.383324647550716
 3.2464293812878067
 5.320117297804017
 3.9971556695645343
 5.039026681750683
 4.709921099884204
 4.93415120764944
 5.131257550693444
 4.955355792546958
 ⋮
 4.227741340484659
 4.523953769215137
 5.2121280936199845
 5.021528576928276
 6.008086254445894
 5.073179235758298
 5.1147852166724865
 5.669970336464331
 5.1045325666422725

---

This notebook was generated using Literate.jl.