Browse Prior Art Database

Practical Function Set for Workforce Management Program

IP.com Disclosure Number: IPCOM000123347D
Original Publication Date: 1998-Sep-01
Included in the Prior Art Database: 2005-Apr-04
Document File: 3 page(s) / 140K

Publishing Venue

IBM

Related People

Gajda, K: AUTHOR [+6]

Abstract

A program is disclosed that generates both a usable and optimal personnel schedule. In order for a personnel scheduler to be usable it must be highly flexible, accommodating company rules, government (federal, state and local) rules, and possibly union rules. In order for the personnel scheduler to be accepted by the employees and management, it must produce schedules that the employee can work, and (within reason) wants to work, and must be entirely fair and explainable in choosing the tasks and times an employee is to work. In order for the personnel scheduler to be effective it must produce nearly optimal schedules (i.e. schedule all important tasks at the best time, by trained employees, and within budget).

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 41% of the total text.

Practical Function Set for Workforce Management Program

   A program is disclosed that generates both a usable and
optimal personnel schedule.  In order for a personnel scheduler to
be usable it must be highly flexible, accommodating company rules,
government (federal, state and local) rules, and possibly union
rules.  In order for the personnel scheduler to be accepted by the
employees and management, it must produce schedules that the
employee can work, and (within reason) wants to work, and must be
entirely fair and explainable in choosing the tasks and times an
employee is to work.  In order for the personnel scheduler to be
effective it must produce nearly optimal schedules (i.e. schedule
all important tasks at the best time, by trained employees, and
within budget).

   The disclosed scheduler has four main components, each of
which has novel and flexible features.
  1.  The forecaster uses data gathered into a database system
      from a point-of-sale system, and produces forecasts for
      key time series as chosen by the customer.  Examples
      include numbers of transactions, dollar totals, item counts
      and so on.  A specially modified Holt-Winters daily
      forecaster is employed.  For a description of the
      Holt-Winters forecaster, see (1).  Day-of-week effects
      are accommodated via the seasonality component in the
      Holt-Winters forecaster.  The special modifications
      include:
      o  handling holiday, first and fifteenth of the month,
         and other special events by de-weighting historical
         data on days when the events occur, applying
         Holt-Winters to the de-weighted data, and re-weighting
         forecast data on days when the events occur.
      o  performing sanity checks based on applying bounds from
         data from 364 days (52 weeks) prior to the forecast
         period.  Other sanity checks, such as those based on
         day-of-week effects, are applied as well.
      o  holiday, first and fifteenth of the month, and other
         special event factors are automatically adjusted when new
         data becomes available.
      o  daily forecasts are turned into quarter hourly forecasts
         in a top-down fashion, using exponential smoothing.  See
         (1) for a description of exponential smoothing.
  2.  The queueing model input consists of the output from the
      forecaster as well as budgetary and/or grade-of-service
      information.  The output is the number of required servers
      for each demand driven task in each quarter hour.  There
      are three possible modes, as chosen by the customer.  In
      the standard queueing mode, the model chosen is based on
      (2).  A second mode computes any customer supplied linear
      combinati...