pick_top_k: Build an optree pipeline that selects up to the top k rows from each group in the given order.

Description

This is an example of building up a desired pre-prepared pipeline fragment from relop nodes.

Usage

pick_top_k(
  source,
  ...,
  partitionby = NULL,
  orderby = NULL,
  reverse = NULL,
  k = 1L,
  order_expression = "row_number()",
  order_column = "row_number",
  keep_order_column = TRUE,
  env = parent.frame()
)

Arguments

source: relop tree or data.frame source.
...: force later arguments to bind by name.
partitionby: partitioning (window function) column names.
orderby: character, ordering (in window function) column names.
reverse: character, reverse ordering (in window function) of these column names.
k: integer, number of rows to limit to in each group.
order_expression: character, command to compute row-order/rank.
order_column: character, column name to write per-group rank in (no ties).
keep_order_column: logical, if TRUE retain the order column in the result.
env: environment to look for values in.

Examples

Run this code


# by hand logistic regression example
scale <- 0.237
d <- mk_td("survey_table",
           c("subjectID", "surveyCategory", "assessmentTotal"))
optree <- d %.>%
  extend(.,
             probability %:=%
               exp(assessmentTotal * scale))  %.>%
  normalize_cols(.,
                 "probability",
                 partitionby = 'subjectID') %.>%
  pick_top_k(.,
             partitionby = 'subjectID',
             orderby = c('probability', 'surveyCategory'),
             reverse = c('probability', 'surveyCategory')) %.>%
  rename_columns(., 'diagnosis' %:=% 'surveyCategory') %.>%
  select_columns(., c('subjectID',
                      'diagnosis',
                      'probability')) %.>%
  orderby(., 'subjectID')
cat(format(optree))

Run the code above in your browser using DataCamp Workspace