对数据帧的相等值/行求和/计数

Question

I have a dataframe that contains a bunch of individual trip data, with a start and end station ID for each trip.

I'm trying to make a second dataframe that is all the info rearranged for each station. So for example, if there are 50 trips with start_station_id == 12 in the first dataframe, then the second dataframe under station_id 12 would have its "starts" column equal to 50

Currently I figured a for loop would be the best method for this, but I cant seem to crack it

for(i in range(station_ids)){
   stationData$starts[i] <- sum(data$start_station_id[i] == station_ids[i])
}

This produces the following error:

Error in `$<-.data.frame`(`*tmp*`, starts, value = c(0, 0, 0, 0, 0, 0,  : 
  replacement has 370 rows, data has 369

station_id's is a variable that contains each unique station id #, stationData$starts is where I want the number of starts stored. Data is the original data I am trying to run the for loop over.

Is there an easier way to complete this operation or am I just writing the for loop wrong? Any tips would be super helpful

Answer 1

From what I have understood from your question you're trying to count the occurences of each station_id this can be easily achieved by the table function which returns a table object ie a named vector containing the count and station_id as the names.

base R

table(data$start_station_id)
data.frame(table(data$start_station_id)) #if you prefer the data.frame look

If you want to have the number of occurence merged into your old data.frame can inner join the two data.frames using the merge function

tbl.df <-data.frame(table(data$start_station_id))
colnames(tbl.df)[1] <- "start_station_id"
data <- merge(data, tbl.df)

data.table

data.table::setDT(data)
data[, `Number of rows` := .N, by = start_station_id]

the := is a data.table function that creates new columns, the .N give the count of the rows for the current group, the by specifies by which column to group. this automatically adds the column Number of rows to the data.table . for an introduction on the data.table package check the vignette . this implementation is the fastest.

Answer 2

The approach is feasible but you should not use a data.frame to put your fresh data in as you don't know the row counts, yet. Just use a list and convert it to data.frame in the end:

stationData = list("station" = unique(c(data$start_station_id, data$end_station_id)), ##  Create a list of all stations
                   "starts" = c(),

for(i in i:length(stationData$station)){
   s = stationData$station[i]
   stationData$starts[i] <- sum(data$start_station_id == s)
}
stationData = as.data.frame(stationData)

At the end of the for-loop the length of both columns will be identical and there will be no problem to create this data.

A much easier way, though, would be to use the table() function, which computes the number of starting stations automatically and was already proposed by Abdessabour Mtk.