In R, categorize time series data based on regex - r

Question

My data is organised as follows: for each product, there is a tax rate for each year as well as a base year tax (baseyr).
product<-c("01","02","03","04")
baseyr <-c("10","8 GBP/tonne","8GBP/tonne + 8GBP/tonne","8")
yr1<-c("5","5 GBP/tonne","5GBP/tonne + 10GBP/tonne","5 + 5GBP/tonne")
yr2<-c("3","3GBP/tonne + 6GBP/tonne","3 GBP/tonne","3 + 5GBP/tonne")
yr3<-c("2","2","2GBP/tonne + 2GBP/tonne","excluded")
sched<-data.frame(product,baseyr,yr1,yr2,yr3)
For each year, I need to classify each product by tax type in a new column based on the following conditions:
#number -> only numbers in the tax
#nonnumber -> numbers and strings in the tax
#mixed -> either two strings or number and string; the two strings are specified by a plus sign
#baseyr -> if the tax is "excluded" from the list, the tax to be used should be the value in base year, and the classification based on this
So if there are 3 years I need to generate 3 tax type columns. However the number of years changes randomly per dataset so I need to code with this in mind. My code is currently something like this:
yearnum<-3 #set number of years; it is between 1 and around 10 but there is no limit
schedule<-c(paste0("yr",1:yearnum)
tax<-c(paste0(schedule,"_tax")
for(i in 1:nrow(sched)){
#for each new tax type
for(j in tax){
#columns 3 to five where the yearly tax rates are
for(yr in 3:5){
#if the tax is excluded from the list, the base year tax should be used to determine the tax nature
if(sched[i,yr] =="excluded"){sched[i,yr] <- sched[i,baseyr]}
#if there is a plus sign it is a mixed tax
if(grepl("\\+",sched[i,yr])){sched[i,j] <- "mixed"}
#if it is not mixed but contains strings it is a nonnumber tax
if(grepl("[:alpha:]",sched[i,yr])){sched[i,j] <- "nonnumber"}
#finally if it is neither of the above it must be a number tax
if(is.na(sched[i,j])){sched[i,j] <- "number"}
}}}
NOTE: I do not know at the start how many years there will be in total; this has to be generated in the code. Any advice much appreciated, especially to avoid these for loops that don't seem to work properly for me.
The final output should be:
#so the output should be:
yr1_tax<-c("number","nonnumber","mixed","mixed")
yr2_tax<-c("number","mixed","nonnumber","mixed")
yr3_tax<-c("number","number","mixed","number")
#and the final dataframe:
sched<-data.frame(product,baseyr,yr1,yr2,yr3,yr1_tax,yr2_tax,yr3_tax)

Answer 1

You could use if_else to change all the excluded into baseyr. Then use case when with regular expressions as shown below:
sched %>%
mutate(
across(starts_with('yr'), ~ifelse(.x == 'excluded', baseyr, .x),
.names = '{.col}_tax'),
across(ends_with('tax'),
~case_when(grepl("^\\d+$", .x) ~ 'number',
grepl('^[^+]$', .x)~'nonnumber',
grepl('[+]', .x)~'mixed')))
product baseyr yr1 yr2 yr3 yr1_tax yr2_tax yr3_tax
1 01 10 5 3 2 number number number
2 02 8 GBP/tonne 5 GBP/tonne 3GBP/tonne + 6GBP/tonne 2 nonnumber mixed number
3 03 8GBP/tonne + 8GBP/tonne 5GBP/tonne + 10GBP/tonne 3 GBP/tonne 2GBP/tonne + 2GBP/tonne mixed nonnumber mixed
4 04 8 5 + 5GBP/tonne 3 + 5GBP/tonne excluded mixed mixed number
The regex is simplified in that I am checking for digits only (number), If there is + then mixed, then if no + then nonnumber.