select_for_report: Select data for report

In jmeydam/countcheck: Detecting Count Anomalies Using Bayesian Hierarchical Models

Select data for report

Description

Selects data for report.

Usage

select_for_report(
  d,
  unit_df = NULL,
  unit_group_name = NULL,
  min_y_new = 0,
  min_diff = 0
)

Arguments

d	Data frame as returned from countcheck()
unit_df	Data frame with columns unit, unit_group_name, unit_name, and unit_url (ignored if NA). Must contain exactly one record for each unit in d. (default: NULL)
unit_group_name	Name of unit group for which data should be returned. Can be part of name (substring). (default: NULL)
min_y_new	Minimum value for y_new (default: 0)
min_diff	Minimum difference between y_new and ucl_partpool (default: 0)

Details

Selected columns:

• unit: ID for unit

• y_new: new count values of interest

• ucl_partpool: UCL based on n_new and partial-pooling estimate of theta

• fe_partpool: factor f, with observed y_new exceeding partial-pooling UCL by f * sd(y_new), given n_new and partial-pooling estimate of theta

Condition that always applies: y_new must exceed ucl_partpool.

Additional optional conditions:

• y_new exceeds min_y_new (default: 0)

• difference between y_new and ucl_partpool exceeds min_diff (default: 0)

• unit_group_name (if present) matches part of group name in unit_df (if present) for units in d

Returned data is sorted by fe_partpool in descending order. Therefore, items towards the top are less likely to be the result of random fluctuations.

Value

Data frame to be passed in countcheck_list to html_report()

jmeydam/countcheck documentation built on Aug. 16, 2024, 11:34 a.m.