GeoTimeseries: Constructs a 'GeoTimeseries' object.

In google/GeoexperimentsResearch: Geo Experiments

Description

Constructs a GeoTimeseries object.

Usage

GeoTimeseries(x, metrics = character(0), date.format = "%Y-%m-%d")

Arguments

x	a plain data.frame or an object that has to be coercible to a plain data.frame with columns date, geo. date must be a Date object or a character vector or factor coercible to Date, and 'geo' must be a character vector, factor, or integer-valued. All columns that start with a dot (.) are removed. If geo is an integer-valued numeric or factor, it is coerced to character silently without error. If date is a character or factor, it is silently coerced to Date. An error is output if the date conversion fails.
metrics	(character) column names that point to numeric columns. At least one metric must be specified. All metrics have to be not all NA.
date.format	(string, optional) format of the column date in the form understood by as.Date(). Used only if the date column is of character type.

Value

A GeoTimeseries object, which is a data.frame, with the required columns,

• date: a vector of class Date.

• geo: a character-valued vector of Geo IDs.

• metrics : one or more numeric metrics. In addition, optionally any other user-definable columns. (These are ignored by the specified methods but may be convenient for the user). Three columns are added, for convenience:

  • .weekday: day number (1=Monday, 2=Tuesday, ..., 7=Sunday)

  • .weeknum: week number in the year from 0 to 53.

  • .weekindex: absolute index of week, year + weeknum (e.g., 201542)

These should be convenient for generating totals and averages, using the aggregate method. The columns 'date' and 'geo' form the primary keys: it is guaranteed that no duplicate ('date', 'geo') pairs exist.

The object includes fields stored in the attribute 'info':

• metrics: names of the metric columns.

• other: names of the other user-supplied columns.

google/GeoexperimentsResearch documentation built on May 17, 2019, 7:42 a.m.