r-functions-performance

This repo was created with the purpose of studying performance of different functions that can be used to obtain the same result.

How it Works

We are using "Melbourne Housing Market" public dataset provided by Kaggle with 60.7k observations and 13 columns to test some functions.

We run the selected function 1000 times over one column using microbenchmark R package and use summary() to analyse the difference in milliseconds between the functions.

Performance Result

load csv functions

Load a .csv file into an R object (data frame/tibble/etc.)

function	times	unit	lq	uq	min	max	median	avg
fread	100	ms	38.250	52.078	37.182	244.789	41.062	48.074
import	100	ms	39.385	55.187	38.124	92.511	46.186	49.661
read.csv	100	ms	508.648	585.146	488.136	1173.685	532.987	570.243
read_csv	100	ms	161.373	191.777	133.286	678.702	173.806	189.244

mapping functions

Apply a specific function to a list or vector.

function	times	unit	lq	uq	min	max	median	avg
lapply	1000	ms	12.205	13.952	11.714	212.913	12.494	14.894
sapply	1000	ms	15.154	17.356	14.404	56.467	15.541	17.657
map	1000	ms	26.620	37.616	25.654	223.574	28.354	33.111
map_chr	1000	ms	27.417	38.271	26.093	225.453	29.188	33.753

simple pattern matching functions

Find a simple pattern in a string, and return the subset

function	times	unit	lq	uq	min	max	median	avg
grep	1000	ms	5.233	5.947	4.837	34.422	5.447	6.096
regexpr	1000	ms	8.224	9.052	7.715	52.361	8.467	9.568
sqldf	1000	ms	118.596	133.969	114.072	461.028	121.131	137.014

subset functions

Subset column in dataframe

function	times	unit	lq	uq	min	max	median	avg
base1	1000	ms	0.017	0.033	0.015	0.105	0.021	0.026
base2	1000	ms	0.017	0.032	0.015	0.203	0.021	0.026
select	1000	ms	1.506	1.637	1.432	33.489	1.554	1.704

to date functions

Convert column to date format

function	times	unit	lq	uq	min	max	median	avg
lubridate	1000	ms	7.370	9.153	6.963	202.276	7.705	9.787
to.date	1000	ms	14.633	16.840	14.011	57.368	15.083	17.058
strptime	1000	ms	212.153	236.698	205.226	464.125	216.453	234.386

to string functions

Convert column as character

function	times	unit	lq	uq	min	max	median	avg
as.character	1000	ms	0.028	0.045	0.006	10.563	0.038	0.048
formatC	1000	ms	13.551	14.481	12.711	62.413	13.822	15.167
paste	1000	ms	14.609	15.289	14.030	207.284	14.801	16.277
toString	1000	ms	16.162	16.837	15.549	58.643	16.381	17.780
sprintf	1000	ms	18.049	18.847	17.456	58.991	18.274	19.786

License

r-functions-performance is MIT Licensed.