ForecastHelpDeskCalls: Forecast the number of calls that will come into a help desk

In stevenranney/ispiranteRanney: Functions to analyze Ispirante data

Description

This function extracts the number of calls received daily at a

help desk and forecasts using an Autoregressive Integrated Moving Average (ARIMA)

model for a number of days into the future.

Usage

ForecastHelpDeskCalls(dF, distance = 31, ...)

Arguments

dF	The dF that needs an weekday value.
distance	How far into the future should the forecast be made? A numeric value passed to the forecast statement that shows the number of calls to the helpdesk in distance number of days. For a monthly forecast, distance = 365/12 is appropriate. distance = 365/4 is appropriate for quarterly predictions. Values could theoretically range from 1 to anything >1. Here, the defaults is to forecast 31 days into the future.
...	Arguments to be passed to other functions. Specifically, this can take the form of identifying which date value is of interest. For example the sample datasets provided all had either at least assigned_at or created_at values. The AssignDateAndDay can used either of these values as the date of interest.

Details

This function calls the AssignDateAndDay function to

assign the day values used. The dayOfWeek values are used

in calculating the number of calls in a given day.

Dates in the dF passed to this function do not need to be

sequential. This function 'fills in' date gaps that are not represented in

the dataFrame and assigns dates that do not appear in the dF a

value of 0.

For the purposes of this proposal, this function only forecasts with

arima though other forecasting models are available. In many

cases, selecting the right model and correct model parameters requires

significant familiarity with the data and would otherwise be left up to the

analyst. Because this process is designed to be automated and simple, a

14-order arima model is used.

arima has a number of arguments that can be changed.

For the purposes of this proposal, the user cannot change these arguments.

The prediction intervals that appear in the subsequent plot from

calling this formula are the defaults for the forecast function.

those defaults are c(80, 95) confidence levels for the prediction

interval.

Author(s)

Steven H. Ranney Contact: [e-mail]

Steven Ranney

See Also

ts

forecast

arima

ggplot

geom_line

stevenranney/ispiranteRanney documentation built on May 30, 2019, 4:46 p.m.