getDTeval

Overview

Using the get() and eval() functions allows for more programmatic coding designs that enable greater flexibility and more dynamic computations. However, in data.table statements, get() and eval() reduce the efficiency of the method by performing work prior to data.table’s optimized computations. getDTeval is useful in translating get() and eval() statements more efficiently for improved runtime performance.

Install the current release from CRAN:

install.packages(‘getDTeval’)

Install the development version from GitHub:

devtools::install_github(‘mb4511/getDTeval’)

Functions

getDTeval package has 2 main functions.

The main purpose of developing the package is to translate get() and eval() statements more efficiently, which allows a user to both incorporate programmatic designs and utilizing data.table’s efficient processing routines.

  1. getDTeval offers a method of fully translating coding statements into an optimized coding statement.

  2. benchmark.getDTeval performs a benchmarking experiment for data.table coding statements that use get() or eval() for programmatic designs. The a) original statement is compared to b) passing the original statement through getDTeval and also to c) an optimized coding statement. The results can demonstrate the overall improvement of using the coding translations offered by getDTeval::getDTeval().

Applications and Benefits

There are some major applications to the getDTeval package:

1). Combining programmatic coding designs with data.table’s efficiency. Better utilizing get() and eval() without the trade-offs in performance.

2). Expanding on the use of eval() in data.table’s calculations.

3). Expanding on the use of eval() in dplyr code.

Examples

Import the data from formulaic package

dat = formulaic::snack.dat

Here the data contains a simulated survey information with records on demographics and other tracked metrics based on the survey responses

names(dat)

[1] “Age”
[2] “Gender”
[3] “Income”
[4] “Region”
[5] “Persona”
[6] “Product”
[7] “Awareness”
[8] “BP_For_Me_0_10”
[9] “BP_Fits_Budget_0_10”
[10] “BP_Tastes_Great_0_10”
[11] “BP_Good_To_Share_0_10”
[12] “BP_Like_Logo_0_10”
[13] “BP_Special_Occasions_0_10” [14] “BP_Everyday_Snack_0_10”
[15] “BP_Healthy_0_10”
[16] “BP_Delicious_0_10”
[17] “BP_Right_Amount_0_10”
[18] “BP_Relaxing_0_10”
[19] “Consideration”
[20] “Consumption”
[21] “Satisfaction”
[22] “Advocacy”
[23] “Age Group”
[24] “Income Group”
[25] “User ID”

Set up some constant names:

mean.age.name = “Mean Age” age.name = “Age” awareness.name = “Awareness” gender.name = “Gender” region.name = “Region” Use cases of benchmark.getDTeval function

sample.dat <- dat[sample(x = 1:.N, size = 10^6, replace = TRUE)] the.statement <- “sample.dat[get(age.name) > 65, .(mean_awareness = mean(get(awareness.name))), keyby = c(eval(gender.name), region.name)]” benchmark.getDTeval(the.statement = the.statement)

category Min. 1st Qu. Median Mean 3rd Qu. Max. getDTeval 0.04582527 0.05954336 0.07194444 0.09330247 0.1003876 0.4570584 optimized statement 0.04378177 0.06732308 0.07925487 0.10189325 0.1004987 0.3444931 original statement 0.13019905 0.17070118 0.18737796 0.20195735 0.2160447 0.4447632 The result shows the reduction in running time using getDTeval over the original statement.

Use cases of getDTeval function

Returning the translated coding statement: the.statement <- “dat[get(gender.name) == ‘Female’, mean(get(age.name)), keyby = region.name]” getDTeval(the.statement = the.statement, return.as = “code”)

[1] “dat[Gender == ‘Female’, mean(Age), keyby = region.name]” Returning the calculation result: getDTeval(the.statement = the.statement, return.as = “result”)

  Region       V1

1: Midwest 54.96774 2: Northeast 55.90385 3: South 55.45205 4: West 54.70430 Returning the a list of the calculation result and the code: getDTeval(the.statement = the.statement, return.as = “all”)

$result Region V1 1: Midwest 54.96774 2: Northeast 55.90385 3: South 55.45205 4: West 54.70430

$code [1] “dat[Gender == ‘Female’, mean(Age), keyby = region.name]” Please check out the vignettes file to see more examples and details.