Important Data Verbs

  1. Create new variables or change/drop existing ones
    1. Can change the cases: summarise(), left_join(), inner_join(), full_join()
    2. Doesn’t change the cases select(), mutate()
  2. No new variables created.
    1. Can change the cases: filter(), semi_join(), anti_join()
    2. Doesn’t change the cases: arrange(), group_by()
  3. gather() and spread()
  1. Create new variables or change existing ones
    1. summarise()
      • arguments should be named. These come the names of the new variables.
      • new_var_name = reduction_verb(...)
      • can use transformation & reduction verbs inside the reduction verb, e.g. stderr = sd(sqrt(var) / n())
      • Names in output will be only the grouping vars of the input and the new variables created.
      • tally() is a shortcut for summarise(n = n())
      • output: one row for each group
    2. joins
      • Output contains all the variables in the two tables.
        • left_join() - there will be at least one case in the output for every case in the left table. Output includes all vars.
        • inner_join() - output has only cases in left table that have a match in right table. Output includes all vars.
        • full_join() - contains all cases from both left and right table. For any cases that don’t have a match in the other table, NA is inserted as the value for the variable for the vars from the other table.
    3. mutate()
      • arguments should be named: these names become the names of the new variables.
      • new_var_name = transformation_verb(...)
      • can use reduction and transformation verbs inside the transformation verbs, e.g. frac =/(count, sum(count))1
      • output: same cases as input but with new variables added.
      • transmute() is similar, but drops all variables other than those created.
    4. select()
      • to rename the selected columns, use named arguments: new_var = old_var
      • rename() works the same but does not throw away any columns
      • output: same cases, but a subset of variables
  2. No new variables created
    1. group_by()
      • no named arguments
      • output: same cases and vars as input but “marked” with groups per the arguments.
    2. arrange() — sometimes with `desc().
      • no named arguments
      • multiple arguments: second and later arguments used to break ties in the arrangement by the previous arguments.
      • respects groups
      • output: same cases and vars as input but in an order specified with arguments
    3. filter()
      • no named arguments: a boolean transformation verb
      • if reduction verbs used, will respect groups.
      • output: same cases and vars as input but some cases may be dropped.
    4. joins, output contains only the left-table variables in the output
      • semi_join() - filters the left-table input to include only those with a match in the right table. Output includes only left-table vars.
      • anti_join() is same as semi_join() but the filter is for those cases without a match in the right table. Output includes only left-table vars.
  3. gather() and spread()

  1. More naturally written as frac = count / sum(count)↩