Approach to Correcting Misspelt Local Government Areas

Motivation

Nigeria has 774 Local Government Areas (LGAs). There are a number of factors that can make working with them challenging:

1. Some LGAs share the same name with their respective States e.g. Bauchi LGA in Bauchi State, and this can lead to wrong application of the data.
2. Some LGAs bear the same name although they are in different States, e.g. Obi LGA in Benue and Nasarawa States.
3. LGA spelling discrepancies are rife in the literature, and believe it or not, even in official government documents.

A proposed solution

The function fix_region has methods that are designed to address spelling errors LGAs¹. These fixes can be carried out in 3 incremental phases; if the user cannot effect the corrections at a particular level s/he can proceed to the next stage. The phases are as follows:

1. Automatic
2. Interactive
3. Manual

1. Automatic fixes

When the spelling error is slight and unambiguous, the function automatically effects the repair. However, when there is only a single misspelt LGA, and especially if it is supplied as a string, fix_region will signal an error.

library(naijR)

fix_region("Legos Island")
#> Error in `fix_region()`:
#> ! Incorrect region name(s); consider reconstructing 'x' with `states()`
#>   or `lgas()` for a more reliable fix

The thinking is that an automated solution may not be necessary for a single value that was probably provided interactively.

When another LGA is added, such that the argument passed to the function is a character vector with more that one element:

fix_region(c("Legos Island", "Amuwo-Odofin"))
#> ℹ Successful fix(es):
#> -------------------
#> * Legos Island => Lagos Island

Sometimes when we use a character vector to perform this check, the function fix_region may find it difficult to decide on what fixes to apply. In such instances, the best thing is to convert the vector into an lgas object.

chars <- c("Legos Island", "Amuwo Odofin")
fix_region(chars)
#> Error in `fix_region()`:
#> ! Incorrect region name(s); consider reconstructing 'x' with `states()`
#>   or `lgas()` for a more reliable fix

# Create an `lgas` object
lgs <- lgas(c("Legos Island", "Amuwo Odofin"))
#> Warning: One or more items is not an LGA. Spelling error?
fix_region(lgs)
#> ℹ Successful fix(es):
#> -------------------
#> * Legos Island => Lagos Island
#> * Amuwo Odofin => Amuwo-Odofin
#> [1] "Lagos Island" "Amuwo-Odofin"
#> attr(,"misspelt")
#> character(0)
#> attr(,"regions.fixed")
#>   Legos Island   Amuwo Odofin 
#> "Lagos Island" "Amuwo-Odofin"

Note that the constructor lgas(), by default, signals a warning when the vector we are supplying has misspelt LGAs. However, to avoid the unnecessary verbosity whilst attempting the fix, this warning is suppressed if lgas() is nested in fix_region().

fix_region(lgas(c("Legos Island", "Amuwo Odofin")))
#> ℹ Successful fix(es):
#> -------------------
#> * Legos Island => Lagos Island
#> * Amuwo Odofin => Amuwo-Odofin
#> [1] "Lagos Island" "Amuwo-Odofin"
#> attr(,"misspelt")
#> character(0)
#> attr(,"regions.fixed")
#>   Legos Island   Amuwo Odofin 
#> "Lagos Island" "Amuwo-Odofin"

When the spelling mistakes depart far from the available LGA spellings, the user is shown a message stating what fixes could not be applied. To now carry out a fix, one can proceed to the next stage.

fix_region(lgas(c("Orange County", "Amuwo Odofin")))
#> ℹ Successful fix(es):
#> -------------------
#> * Amuwo Odofin => Amuwo-Odofin
#> 
#> Fix(es) not applied:
#> --------------------
#> * Orange County
#> [1] "Orange County" "Amuwo-Odofin" 
#> attr(,"misspelt")
#> [1] "Orange County"
#> attr(,"regions.fixed")
#>   Amuwo Odofin 
#> "Amuwo-Odofin"

2. Interactive fixes

When the automatic fix is not feasible, the user has the option of doing it interactively by calling fix_region and setting its interactive argument to TRUE. By following the prompts, the misspelt LGA as well as possible replacements are presented. All the user needs to do is to select the desired replacement value. This is particularly useful when the user is not sure of what the correct spelling might be.

When a misspelt LGA has more than one match, the interactive approach is the viable option for effecting fixes. In the example below, not all the mispelt LGAs could be fixed automatically:

mispelt.adamawa <- c("Fufure", "Demsa", "Machika", "Fufure", "Ganye", "Hong")

# check for misspelt LGAs and, if necessary, attempt to fix
length(mispelt.adamawa)
#> [1] 6
sum(is_lga(mispelt.adamawa))
#> [1] 3
corrected.adamawa <- fix_region(mispelt.adamawa)
#> 'Machika' approximately matched more than one region - Michika, Machina
#> ℹ Successful fix(es):
#> -------------------
#> * Fufure => Fufore
#> 
#> Fix(es) not applied:
#> --------------------
#> * Machika
sum(is_lga(corrected.adamawa))
#> [1] 5

We see that the string ‘Machika’ is not an actual LGA and there are more than one possible candidate LGA that are considered as possible replacement. Our original intent was to use “Michika LGA”. To address this mistake, we can run fix_region interactively. Note that the interactive option is only available for use with objects of class lgas.

adamawa <- fix_region(lgas(corrected.adamawa), interactive = TRUE)

Next, the user is provided with a prompt to provide a search item for likely LGAs that would be an appropriate replacement for the misspelt one. Possible replacements are presented and to select any value, the appropriate number should be entered at the prompt. The option RETRY allows the user to restart the query and provide another search term, SKIP is to try out a different misspelt item (where there are more than one) and to QUIT is to exit the prompt. On Windows machines, the user can use the native dialog boxes by setting graphic to TRUE.

# Confirm that the spelling mistakes have been fixed.
all(is_lga(adamawa))
#> [1] TRUE

3. Manual fixes

When spelling errors are identified and the correct one is already known, then a manual fix can be applied. For this, we use the function fix_region_manual and its use is straightforward²:

adamawa <- fix_region_manual(corrected.adamawa,
                             wrong = "Machika",
                             correct = "Michika")
adamawa
#> [1] "Fufore"  "Demsa"   "Michika" "Fufore"  "Ganye"   "Hong"