aw_anomaly_report: Anomaly Report

View source: R/anomaly_report.R

aw_anomaly_reportR Documentation

Anomaly Report

Description

Get an anomaly report for one or more metrics

Usage

aw_anomaly_report(
  company_id = Sys.getenv("AW_COMPANY_ID"),
  rsid = Sys.getenv("AW_REPORTSUITE_ID"),
  date_range = c(Sys.Date() - 31, Sys.Date() - 1),
  metrics,
  granularity = "day",
  segmentId = NA,
  quickView = FALSE,
  anomalyDetection = TRUE,
  countRepeatInstances = TRUE,
  debug = FALSE
)

Arguments

company_id

Company Id. Taken from the global environment by default if not provided.

rsid

Adobe report number

date_range

A two length vector of start and end Date objects (default set to show last 30 days)

metrics

Metric to request the anomaly detection. If multiple metrics, each metric and date will have it's own row.

granularity

Use either hour, day (default), week, or month

segmentId

Use segments to globally filter the results. Use 1 or many.

quickView

Return a list of 3 lists per metric. 1. All Data 2. Data filtered to include only anomalous rows 3. Interactive ggplot line graph

anomalyDetection

logical statement for including anomaly. Default is TRUE

countRepeatInstances

Should the data include repeat instances

debug

default is FALSE but set to TRUE to see the json request being sent to the Adobe API

Value

If quickView = 'FALSE' (default) then a data frame including the day, metric, data, dataExpected, dataUpperBound, dataLowerBound, and dataAnomalyDetected will be returned. If quickView = 'TRUE' then a list of three lists will be returned. The first list will be a data frame including all the default columns. The second list item will be a filtered data frame that includes rows where dataAnomalyDetected = 'TRUE'. The third list item is a visual made using 'ggplot2' with the error band and points where the dataAnomalyDetected = 'TRUE'. If more than one metric is in the request and quickView is set to TRUE then the lists will be split by each metric requested.


adobeanalyticsr documentation built on Nov. 9, 2023, 5:07 p.m.