Matt Gregory bio photo

Matt Gregory

Data Scientist

Twitter Github

Overview

Character matching provides a powerful way to make lookup tables. There are more concise functions available in packages like dplyr that achieve the same end but it is useful to understand how they are implemented with basic subsetting.

library(dplyr)

We start off by building an example dataframe.

set.seed(1337)  # we use rnorm

pupil_data <- data.frame(
studentid = 1:12,
school = c("Park view", "Grange Hill", "Sweet valley"),
superoutputarea = c("E01011949", "E01011105", "E01011333"),
attainment = (rnorm(n = 12)),
stringsAsFactors = FALSE # we can modify the class of variables later if required
)

If we look at the data, we notice the variable superoutputarea is a nine digit code that doesn’t tell a human much. We are interested in how the area relates to the socio-economic classification of typical people who live in that area or a measure of deprivation of the area. We must convert this into the more informative proxy which can then be used in our machine learning tools later.

pupil_data
##    studentid       school superoutputarea attainment
## 1 1 Park view E01011949 0.1924919
## 2 2 Grange Hill E01011105 -1.4467018
## 3 3 Sweet valley E01011333 -0.3231805
## 4 4 Park view E01011949 1.6222961
## 5 5 Grange Hill E01011105 -0.6890241
## 6 6 Sweet valley E01011333 2.0421222
## 7 7 Park view E01011949 0.9437791
## 8 8 Grange Hill E01011105 2.0819269
## 9 9 Sweet valley E01011333 1.9171173
## 10 10 Park view E01011949 -0.4148122
## 11 11 Grange Hill E01011105 1.0328535
## 12 12 Sweet valley E01011333 -1.6785696
str(pupil_data)
## 'data.frame':	12 obs. of  4 variables:
## $ studentid : int 1 2 3 4 5 6 7 8 9 10 ...
## $ school : chr "Park view" "Grange Hill" "Sweet valley" "Park view" ...
## $ superoutputarea: chr "E01011949" "E01011105" "E01011333" "E01011949" ...
## $ attainment : num 0.192 -1.447 -0.323 1.622 -0.689 ...

What we need is a list which contains the necessary translation for superoutputarea. We define that here as lookup. A 7-point scale is used for deprivation with Sweet Valley High in a wealthy area and Park View in a deprived area.

lookup <- c("E01011949" = 1, "E01011105" = 3, "E01011333" = 7)

To convert we simply:

lookup[pupil_data$superoutputarea]
## E01011949 E01011105 E01011333 E01011949 E01011105 E01011333 E01011949 
## 1 3 7 1 3 7 1
## E01011105 E01011333 E01011949 E01011105 E01011333
## 3 7 1 3 7
#  if we don't want the names in the result
unname(lookup[pupil_data$superoutputarea])
##  [1] 1 3 7 1 3 7 1 3 7 1 3 7

Thus we can use this to create a new variable called depriv.

pupil_data$depriv <- NULL
pupil_data$depriv <- unname(lookup[pupil_data$superoutputarea])
pupil_data
##    studentid       school superoutputarea attainment depriv
## 1 1 Park view E01011949 0.1924919 1
## 2 2 Grange Hill E01011105 -1.4467018 3
## 3 3 Sweet valley E01011333 -0.3231805 7
## 4 4 Park view E01011949 1.6222961 1
## 5 5 Grange Hill E01011105 -0.6890241 3
## 6 6 Sweet valley E01011333 2.0421222 7
## 7 7 Park view E01011949 0.9437791 1
## 8 8 Grange Hill E01011105 2.0819269 3
## 9 9 Sweet valley E01011333 1.9171173 7
## 10 10 Park view E01011949 -0.4148122 1
## 11 11 Grange Hill E01011105 1.0328535 3
## 12 12 Sweet valley E01011333 -1.6785696 7
str(pupil_data)  #  check variables are of appropriate class
## 'data.frame':	12 obs. of  5 variables:
## $ studentid : int 1 2 3 4 5 6 7 8 9 10 ...
## $ school : chr "Park view" "Grange Hill" "Sweet valley" "Park view" ...
## $ superoutputarea: chr "E01011949" "E01011105" "E01011333" "E01011949" ...
## $ attainment : num 0.192 -1.447 -0.323 1.622 -0.689 ...
## $ depriv : num 1 3 7 1 3 7 1 3 7 1 ...

Great, now we can use this dataframe for machine learning. What if we have a large dataframe, are there more concise and faster ready made functions to use? Probably but we won’t elucidate that here, we just assume dplyr is fast as it passess to C++. Plus I like dplyr with its nice chaining.

mutate(pupil_data, depriv2 = unname(lookup[pupil_data$superoutputarea]))
##    studentid       school superoutputarea attainment depriv depriv2
## 1 1 Park view E01011949 0.1924919 1 1
## 2 2 Grange Hill E01011105 -1.4467018 3 3
## 3 3 Sweet valley E01011333 -0.3231805 7 7
## 4 4 Park view E01011949 1.6222961 1 1
## 5 5 Grange Hill E01011105 -0.6890241 3 3
## 6 6 Sweet valley E01011333 2.0421222 7 7
## 7 7 Park view E01011949 0.9437791 1 1
## 8 8 Grange Hill E01011105 2.0819269 3 3
## 9 9 Sweet valley E01011333 1.9171173 7 7
## 10 10 Park view E01011949 -0.4148122 1 1
## 11 11 Grange Hill E01011105 1.0328535 3 3
## 12 12 Sweet valley E01011333 -1.6785696 7 7

Sometimes we might have a more complicated lookup table which has multiple columns of infomation. Suppose we take our vector of attainment grades and round them to the nearest whole number.

pupil_data <- mutate(pupil_data, grade = round(attainment, digits = 0))

grades <- pupil_data$grade

info <- data.frame(
grade = -3:3,
desc = c("Awful", "Rubbish", "Poor", "OK", "Satisfactory", "Good", "Awesome"),
pass = c(F, F, F, T, T, T, T)
)

We want to duplicate the info table so that we have a row for each values in grade. We can do this in two ways either using match() and integer subsetting, or rownames() and character subsetting:

# using match
id <- match(grades, info$grade)
info[id, ]
##     grade         desc  pass
## 4 0 OK TRUE
## 3 -1 Poor FALSE
## 4.1 0 OK TRUE
## 6 2 Good TRUE
## 3.1 -1 Poor FALSE
## 6.1 2 Good TRUE
## 5 1 Satisfactory TRUE
## 6.2 2 Good TRUE
## 6.3 2 Good TRUE
## 4.2 0 OK TRUE
## 5.1 1 Satisfactory TRUE
## 2 -2 Rubbish FALSE
# using rownames
rownames(info) <- info$grade
info[as.character(grades), ]
##      grade         desc  pass
## 0 0 OK TRUE
## -1 -1 Poor FALSE
## 0.1 0 OK TRUE
## 2 2 Good TRUE
## -1.1 -1 Poor FALSE
## 2.1 2 Good TRUE
## 1 1 Satisfactory TRUE
## 2.2 2 Good TRUE
## 2.3 2 Good TRUE
## 0.2 0 OK TRUE
## 1.1 1 Satisfactory TRUE
## -2 -2 Rubbish FALSE

We have matched the grade of the student with its appropriate descriptor and pass / fail status using a more complicated lookup table.

Conclusion

A named character vector can act as a simple lookup table. We could even read this in from a csv file. Lookup is simple in R.

References

  • Wickham, H. (2015). Advanced R.
sessionInfo()
## R version 3.2.3 (2015-12-10)
## Platform: x86_64-pc-linux-gnu (64-bit)
## Running under: Ubuntu 14.04.4 LTS
##
## locale:
## [1] LC_CTYPE=en_GB.UTF-8 LC_NUMERIC=C
## [3] LC_TIME=en_GB.UTF-8 LC_COLLATE=en_GB.UTF-8
## [5] LC_MONETARY=en_GB.UTF-8 LC_MESSAGES=en_GB.UTF-8
## [7] LC_PAPER=en_GB.UTF-8 LC_NAME=C
## [9] LC_ADDRESS=C LC_TELEPHONE=C
## [11] LC_MEASUREMENT=en_GB.UTF-8 LC_IDENTIFICATION=C
##
## attached base packages:
## [1] methods stats graphics grDevices utils datasets base
##
## other attached packages:
## [1] dplyr_0.4.3 testthat_0.11.0 knitr_1.12.3
##
## loaded via a namespace (and not attached):
## [1] Rcpp_0.12.2 digest_0.6.8 crayon_1.3.1 assertthat_0.1
## [5] R6_2.1.1 DBI_0.3.1 formatR_1.2.1 magrittr_1.5
## [9] evaluate_0.8 stringi_1.0-1 lazyeval_0.1.10 tools_3.2.3
## [13] stringr_1.0.0 parallel_3.2.3 memoise_0.2.1