Getting summaries by subgroups of your data {plyr}

• highest profit margin (as point data)

highestMargin <- max(companiesData$margin)

highestMargin <- companiesData[companiesData$margin == max(companiesData$margin),]

highestMargin <- subset(companiesData, margin==max(margin))

• the highest margin (as dataframe)

  • NULL as the second argument for factors to split by

highestProfitMargin <- ddply(companiesData, NULL, summarize, bestMargin = max(margin))

• the highest profit margin for each company? (add new column(s))

myResults <- ddply(companiesData, 'company', transform, highestMargin = max(margin), lowestMargin = min(margin))

highestProfitMargins <- ddply(companiesData, 'company', summarize, bestMargin = max(margin))

highestProfitMargins <- ddply(companiesData, 'company', transform, bestMargin = max(margin))

• just the (entire) rows that have the highest profit margins.

highestProfitMargins <- ddply(companiesData, 'company', function(x) x[x$margin==max(x$margin),])

• plyr::ddplr(): Input a data frame and get a data frame back.

ddply(mydata, c('column name of a factor to group by', 'column name of the second factor to group by'), summarize OR transform, newcolumn = myfunction(column name(s) I want the function to act upon))

• the first argument
  • the name of the original data frame
• the second argument
  • the name of the column or columns you want to subset your data by.
    • No parentheses are needed for just one factor if you're using quotation marks
    • NULL
      • select the row with highest margin in the entire data set
• The third argument
  • tells ddply() whether to return
    • summarize
      • doesn't give any information from other columns in the original data frame
    • transform
      • return your existing data frame with a new column
• the fourth argument
  • names the new column and then
  • lists the function you want ddply() to use.

• highest profit margin

highestMargin <- max(companiesData$margin)

highestMargin <- companiesData[companiesData$margin == max(companiesData$margin),]

highestMargin <- subset(companiesData, margin==max(margin))

• the highest profit margin for each company?

  • applying a function by groups
    • what R calls factors.
  • "split-apply-combine" (Hadley Wickham {plyr})
    • Split up your data set by one or more factors,
    • apply some function,
    • then combine the results back into a data set.
  • plyr package created by Hadley Wickham
    • group of "ply" functions
      • ddply(): Input a data frame and get a data frame back.
      • alply(): input an array and get back a list
      • ldply(): input a list and get back a data frame

• install & load

  install.packages("plyr")
  library("plyr")
  

• format of plyr::ddplr() 1: names in quotes

ddply(mydata, c('column name of a factor to group by', 'column name of the second factor to group by'), summarize OR transform, newcolumn = myfunction(column name(s) I want the function to act upon))

• the first argument
  • the name of the original data frame
• the second argument
  • the name of the column or columns you want to subset your data by.
    • No parentheses are needed for just one factor if you're using quotation marks
    • NULL
      • select the row with highest margin in the entire data set
• The third argument
  • tells ddply() whether to return
    • summarize
      • doesn't give any information from other columns in the original data frame
    • transform
      • return your existing data frame with a new column
• the fourth argument
  • names the new column and then
  • lists the function you want ddply() to use.

highestProfitMargins <- ddply(companiesData, 'company', summarize, bestMargin = max(margin))

highestProfitMargins <- ddply(companiesData, 'company', transform, bestMargin = max(margin))

• apply more than one function at a time

myResults <- ddply(companiesData, 'company', transform, highestMargin = max(margin), lowestMargin = min(margin))

• just the (entire) rows that have the highest profit margins.

highestProfitMargins <- ddply(companiesData, 'company', function(x) x[x$margin==max(x$margin),])

• function(x)

  • an anonymous (unnamed, ad-hoc 特別指定) function is coming next

    • extracting a subset of x
      • x refers to the data frame that was passed into the anonymous function
    • the anonymous function is being passed into a ddply() statement that's splitting the data frame by company, what's returned is the matching row(s) for each company.

  • [x$margin==max(x$margin),]

    • I want to match every row where x$margin equals the maximum of x$margin.
    • The comma after x$margin==max(x$margin)
      • tells R to return every column of those matching rows, since no columns were specified.
    • As an alternative, we could seek to return only one or several of the columns instead of all of them.

companiesData[companiesData$margin==max(companiesData$margin)

   - alone, without ddply\(\), gives the highest overall margin, not the highest margin for each company.

• the highest margin in the entire data set

  • NULL as the second argument for factors to split by

highestProfitMargin <- ddply(companiesData, NULL, summarize, bestMargin = max(margin))

• format of plyr::ddplr() 2: dot before the column names

myresult <- ddply(mydata, .(column name of factor I'm splitting by, column name second factor I'm splitting by), summarize OR transform, newcolumn = myfunction(column name I want the function to act upon))

• To get the highest profit margins for each company
  • splitting the data frame by only one factor
    • company.
      • Even if you've only got one factor,
        • it needs to be in parentheses after that dot if you're using the dot

highestProfitMargins <- ddply(companiesData, .(company), summarize, bestMargin = max(margin))

Reference:

• 4 data wrangling tasks in R for advanced beginners | Computerworld