WO2016025986A1

WO2016025986A1 - Roster design methods and systems

Info

Publication number: WO2016025986A1
Application number: PCT/AU2015/000498
Authority: WO
Inventors: Lisa SPIDEN; Michael Pollitt; Armin ARDEKANI; Steven MAIN; Heng-Soon GAN
Original assignee: Roster Right Pty Ltd
Priority date: 2014-08-19
Filing date: 2015-08-19
Publication date: 2016-02-25
Also published as: AU2017100329A4; AU2015306072A1

Abstract

The relates to a computer-implemented method for designing a roster composition for a business, the roster comprising a plurality of tasks and the business employing a plurality of staff types. The method includes determining one or more rules for the roster, at least one rule defining minimum coverage requirements for the tasks, creating a digital representation of the rules, and deriving, from the digital representation, an optimal staff composition that provides a minimum labour cost to satisfy all the rules.

Description

ROSTER DESIGN METHODS AND SYSTEMS

FIELD OF THE INVENTION

[0001] The present invention relates to computerised methods and systems for designing and/or optimising a staff roster for a business. The present invention has broad application to all industries.

INCORPORATION BY REFERENCE

[0002] This application claims priority from Australian provisional patent application no 2014903249 filed 19 August 2014. The complete contents of that application are incorporated herein by reference.

BACKGROUND OF THE INVENTION

[0003] Managing staff rosters is a challenge for any business. On the one hand, it is important to ensure that adequate staff coverage is provided to meet minimum coverage requirements at any given time during the day, week or month. Furthermore, there are numerous legal requirements that must be adhered to which define the terms and conditions of employment for each employee as defined by legislation. In Australia these legal requirements take the form of National Employment Standards which underpin relevant industry specific Awards. Industry Awards may be amended in the form of enterprise agreements between the business and its staff but which must provide a net improvement of terms to employees and must also adhere to the minimum requirement of the National Employment Standards. These legal requirements define, but are not limited to, minimum or maximum hours per week, minimum and maximum shift lengths, and the provision of mandatory breaks. Furthermore, a business will also have internal

requirements to accommodate. Furthermore a business may have specific employment conditions associated with specific employees. Accordingly, ensuring that a roster meets all of these obligations can be a difficult task on its own.

[0004] However, merely ensuring that a roster meets these minimum staffing requirements does not mean that the resources are being allocated efficiently. Inefficiencies may be incurred. An inefficient roster may, for example, provide excessive coverage at different times, in order to meet minimum and maximum shift lengths and to ensure that adequate breaks are provided as defined by legislation.

[0005] Furthermore, it will also be appreciated that different employees are paid at different rates appropriate to the complexity of the employee's role, and their seniority and/or experience, and penalties/or allowances. Minimum rate standards are defined by legislation, including but not limited to an award or enterprise agreement (in Australia), or may be agreed individually with an employee. Therefore, if more expensive staff are rostered too frequently or for too many hours, the business will incur higher staff costs. Similarly, costs can be driven up if the roster results in excessive overtime payments where the rostered staffing levels are inappropriate. All of this, in turn, drives down the profitability of the business. Accordingly, generating a roster that makes efficient use of staff is an important aspect of any business.

[0006] Traditionally, a staff roster is created manually (although usually with the aid of computer visualisation) by a staff manager who will factor in people and availability with little regard to optimisation, productivity or cost outcomes. A staff manager may (i) determine peak periods of work, (ii) assign management to key periods when they are required and sales demand is high, (iii), assign full-time staff, (iv) assign part-time staff, (v) assign casual staff, (vi) make adjustments to ensure adequate coverage is provided, and necessary breaks and days on/days off are provided, and finally (vii) look for ways to reduce overtime/costs.

[0007] This traditional process is labour-intensive. Furthermore, there is no guarantee that the resulting roster will be an efficient use of staff resources - in fact, the opposite is likely to be true. Manual rostering also often inadvertently leads to breaking legal requirements due to the number of rules which all must be adhered to and the complexities of the interaction between these rules.

[0008] Accordingly, there is a need for improved methods and systems to provide a more efficient way to design and/or optimise a roster for a business, to improve the resulting allocation of staff in a roster, to ensure the roster complies with the governing legal obligations, or to at least provide a commercial alternative to current methods and systems for designing a roster which may provide improvements in the efficiency of staff allocation and therefore reduced business costs. SUMMARY OF THE INVENTION

[0009] According to an aspect of the present invention, there is provided a computer- implemented method for designing a roster composition for a business, the roster comprising a plurality of tasks and the business employing a plurality of staff types, the method comprising:

(a) determining one or more rules for the roster, at least one rule defining minimum coverage requirements for the tasks;

(b) creating a digital representation of the rules;

(c) deriving, from the digital representation, an optimal staff composition that provides a minimum labour cost to satisfy all the rules.

[0010] The present invention aims to find a balance between minimising operating costs and adhering to legislative and business requirements which govern how staff can be rostered.

[0011] The rules may include fixed business rules which define the operational framework of the business, and also define flexible business rules. The rules may further include legal rules specifying the legal obligations of the business, for instance under legislation, industry awards, or business-specific enterprise agreements. The present invention can therefore be used to ensure compliance with these rules (legal rules, fixed business rules and flexible rules), while also identifying the minimum labour cost (within the practical constraints of available computing power) of a roster.

[0012] The rules may be determined in a structured, scientific way. In particular, the method may include prompting a user (such as a representative of the business) to specify the legal rules they operate under (e.g. which industry award) or to specify the enterprise agreement they have entered into. The user may also be prompted to enter various fixed and flexible business rules. The computer-implemented method may include receiving, from a user, input describing these rules. Coverage requirements may be explicitly input by a user, or may be determined in other ways - for example, by analysing a current roster provided by the business. [0013] Advantages that may be achieved, from preferred embodiments of the present invention, may include one or more of :

(a) ensuring business compliance with legal rules governing employment of staff;

(b) interrogating, challenging, and validating assumptions around current coverage requirements, and then to provide a new optimised set of minimum coverage requirements in order to meet legal rules and business rules;

(c) provide insight and cost analysis of business decisions relating to labour deployment;

(d) to minimise the cost of labour requirements; and

(e) reduce the cost and time of designing/optimising rosters.

[0014] The roster composition does not necessarily allocate specific staff members to defined shifts. Instead, it may allocate 'staff types' to coverage requirements. This enables the optimisation of the number and composition of staff types in order to meet minimum coverage requirements, thereby meeting the minimum business and legal requirements. This roster composition can then be used as a baseline minimum roster cost to meet the coverage requirements and other business rules. Rosters (actual staff in actual shifts) may, in some cases, be generated automatically from the roster composition baseline. However, in other cases this may be left to be performed manually by a user (e.g. business manager), to allow the user to make discretionary choices regarding specific staff members.

[0015] The present invention deals with staff types, rather than individual staff members. However, in some cases a business may have individual agreements with particular staff members, and there may be rules specific to a staff member which reflect that individual agreement. The present invention therefore allows the application of such rules as being applicable to a single member of a staff type. Having generated the optimal roster composition, a staff manager may then apply this baseline to an actual roster where rules applicable to a single member of a staff type are highlighted as a mandatory roster allocation of the applicable staff member. [0016] The legislative and award requirements, in particular, are complex and regularly subject to change, and it can be very difficult for employers to otherwise ensure their legal obligations are fully complied with.

[0017] The rules may also include fixed business rules. These may define the operational state of the business. Fixed business rules in a retail example may include the number of stores, the size of each store, the opening hours, the location and turnover of each store.

[0018] The rules may also include workforce planning rules. Workforce planning rules may include task allocations for employee types, employee productivity measures, or foot traffic counters within a store (for a retail example). These rules may also include associations between weather forecasts, sales data, task allocation, call volumes (in a call centre environment) to map and forecast staff requirements. These allow the business to plan their workforce requirements.

[0019] Furthermore, the rules may include flexible business rules. These include business-specific rules that can be chosen and varied by the business. For example, they may include the staff types that may perform a particular task - some tasks may require a particular level of seniority, or may require staff who are specially trained to perform the task. Other flexible business rules may specify a minimum staff coverage required at a particular time of day. Other flexible business rules may specify preferred shift lengths for a staff type.

[0020] These rules may be formulated as a digital computer representation, comprising a plurality of variables with constraints that, indirectly or directly, affect the allocation of staff types to tasks. This digital representation may be created dynamically or interactively, or some aspects may be hard-coded. Once the digital representation is created, the step of calculating the roster composition may be performed by determining the value of these variables such that the cost is minimised without violating the defined constraints (i.e. complying with the defined rules). This may be performed by a

commercial evaluation/optimisation engine, or a customised solving engine could be produced. [0021] In addition, in some embodiments, a cost may be calculated for each defined rule. This will allow a business to understand the real cost of each rule, enabling the business to make informed decisions about the application of each rule, allowing greater control of their operating practices and the corresponding labour costs.

[0022] Deriving the roster composition may include providing the digital

representation to a solving engine. It may also include solving the problem of determining a minimum roster cost with a modular approach, including providing digital

representations of one or more sub-problems of the roster problem to the solving engine. The optimised solutions to the sub-problem(s) may be fed back into the solving engine to solve the parent rostering problem.

[0023] According to a further aspect of the present invention, there is provided a method of calculating a cost associated with one or more rostering requirements of a business, the method including:

(a) calculating a first optimised cost for a roster composition with a first set of rules;

(b) calculating a second optimised cost for a roster composition with a second set of rules, the second set of rules having at least one differing rule which differs from the first set of rules, whereby the difference between the first optimised cost and second optimised cost indicates a cost of the at least one differing rule.

[0024] According to a further aspect of the present invention, there is provided a computer-implemented method for designing a roster composition for a plurality of staff members over a roster time period, each staff member having a staff type, the method comprising:

(a) identifying a plurality of tasks to be performed over the roster time period, each task being mapped to one or more permitted staff types;

(b) defining a plurality of rules governing the allocation of staff types to tasks, (c) producing a digital representation of the rules, comprising a plurality of variables each defining an aspect of the roster composition and a range of permitted values for each variable to thereby impose constraints on the roster composition;

(d) determining a cost function for at least some of the variables;

(e) determining desired values of the variables to minimise the sum of the costs, within the permitted ranges of the rules; and

(f) translating the desired values to produce a roster composition corresponding to the desired values.

[0025] In further aspects of the present invention, there are provided methods for costing a roster composition, and for checking feasibility of a roster composition. For example, in one aspect, there is provided a computer-implemented method for checking the compliance of a roster composition with a plurality of roster rules, the roster composition allocating staff types to tasks over a roster time period, the method comprising:

(a) creating a digital representation of the rules, having a number of variables and having constraints for values of the variables;

(b) mapping the roster composition to the variables, to assign variable values to the variables; and

(c) checking whether the variable values are within the constraints.

[0026] According to a further aspect of the present invention, there is provided a computer-implemented system for designing a roster composition, costing a roster composition, or checking feasibility of a roster composition, said system comprising one or more computers including: at least one processor; and at least one storage medium operatively coupled to said processor, said storage medium containing program instructions for execution by said processor, said program instructions causing said processor to execute the steps of the above method.

[0027] According to a still further aspect of the present invention, there is provided a tangible computer-readable medium having computer-executable instructions stored thereon for directing a programmable device to perform the above method, or any of the computer-implemented methods described herein.

[0028] The invention is herein described as 'optimising' a legally compliant roster composition to determine a 'minimum' cost. These terms must be understood in the context of the present invention, which can deal with extremely complex problems. Accordingly, it will be understood that these terms are not used to refer to absolute 'optimum' roster compositions having an absolute 'minimum' cost. For some businesses having particularly complex rostering requirements, it may not be possible to search the entire problem space for all possible solutions. Therefore, due to computer resourcing requirements, the present invention may not always guarantee that the actual roster composition is 'optimised' for 'minimum' cost, only that it is optimised to produce the minimum cost given the depth of the problem search that was able to be performed within the available time. It should also be noted that, in some cases, the 'cost' may be calculated or weighted in a way such that it is different to the strict monetary cost.

[0029] A detailed description of one or more embodiments of the invention is provided below, along with accompanying figures that illustrate by way of example the principles of the invention. While the invention is described in connection with such embodiments, it should be understood that the invention is not limited to any embodiment. On the contrary, the invention encompasses numerous alternatives, modifications and equivalents.

[0030] For the purpose of example, numerous specific details are set forth in the following description in order to provide a thorough understanding of the present invention. The present invention may be practiced according to the claims without some or all of these specific details. For the purpose of clarity, technical material that is known in the technical fields related to the invention has not been described in detail so that the present invention is not unnecessarily obscured.

BRIEF DESCRIPTION OF DRAWINGS

[0031] Figure 1 is a partial screenshot from a user interface for an embodiment of the present invention, displaying a list of tasks;

[0032] Figure 2 is a partial screenshot from a user interface for an embodiment of the present invention, displaying coverage requirements for particular tasks;

[0033] Figure 3 is a partial screenshot showing the mapping of staff types to tasks;

[0034] Figure 4 is a table showing coverage requirements for a roster composition;

[0035] Figure 5 depicts a simple example of work and break segments;

[0036] Figure 6 depicts a table showing cost calculations broken down into cost components;

[0037] Figure 7 graphically depicts a roster composition generated according to an embodiment of the present invention;

[0038] Figure 8 graphically depicts the coverage provided by the roster composition of Figure 7;

[0039] Figure 9 is a schematic diagram depicting an alternative embodiment of the present invention; and

[0040] Figure 10 schematically depicts the translation of engine output into a roster composition..

DETAILED DESCRIPTION Hardware

[0041] The present invention relates to a computer- implemented method and a computer system. It should be appreciated that the hardware used to implement the method of the invention may be conventional in nature or specifically designed for the purpose. As an example of typical hardware, it may include at least one processor or central processing unit (CPU). The CPUs may be interconnnected via a system bus to a random access memory (RAM), read-only memory (ROM), input/output (I/O) interface for connection to input devices (such as a keyboard or mouse) or output devices (such as a screen or speaker), a communications interface for connecting the system to a data processing network (such as the Internet, an Intranet, or a local area network (LAN)).

Overall Scope of the System

[0042] The high-level, overall approach facilitated by a preferred embodiment of the present invention is to perform the following:

(a) review the current roster for feasibility and cost;

(b) define the labour requirements in the form of coverage requirements;

(c) define the other rules (e.g. legal, fixed business rules, or flexible business rules) that should be applied to these coverage requirements;

(d) create a solvable mathematical problem from these rules and requirements which will, as its solution, provide the lowest cost roster without breaking any of constraints;

(e) provide result sets with metrics and data to review and cost individual business decisions; and

(f) determine cost and feasibility of user-input rosters. Roster Structures

[0043] Creating a roster, therefore, begins with the definition of coverage

requirements. These coverage requirements form part of the fixed business rules and define, but are not limited to, parameters such as store or site opening and closing hours.

[0044] A roster coverage requirement can be comprised of tasks. In the retail industry, there may be a limited number of simply defined tasks - such as general sales, store opening, store closing, management meetings (for store managers on a regular basis) and order unpacking (for unpacking deliveries). Figure 1 depicts a partial screenshot from a user interface which depicts a list of different tasks that may need to be rostered.

[0045] Each task may have further defined properties, including the day, the start time, the end time, and the minimum coverage requirements (i.e. number of people who should be assigned to the task). Store opening and closing tasks, for example, may only require a single staff member to complete, whereas general sales roles in peak periods may require larger numbers of staff to be rostered on for that task. Figure 2 depicts a partial screenshot from a user interface which provides an example of defined start and end times for a store opening task.

[0046] Each daily roster may have a number of tasks set in defined blocks of any length throughout the day, and the duration of the roster period may be set by examining the start time of the first task and the end time of the final task defined. Tasks may be sequential or non sequential and may also overlap within a single store. Tasks may be fixed, in that they have a defined start and end time which must not be altered. We may also have floating tasks which may be performed at any time of the day during the opening hours.

[0047] Staff types may also be defined for the business. Staff may be categorised by staff role; for example in an applicable award roles are defined as Employee Level 1 through Employee Level 8, or in the General Retail Industry Award 2010 in Australia, applicable enterprise agreement business defined names may be applied which are mapped to these award levels such as store manager, assistant manager, and shop assistant. Staff may also be categorized by employment type - i.e. full-time, part-time or casual. Each possible combination of staff role and employment type will constitute a different staff type for the purposes of this embodiment of the present invention. In some embodiments, each individual staff member could even be assigned their own 'staff type', although for many applications this level of resolution is unnecessary and may not be desired. Instead, a user could be enabled to allocate specific staff members to staff types in the final roster composition, to generate the final roster for the business.

[0048] The number of different applicable staff roles and employment types will depend on the business structure, and the present invention may be employed for any number of staff types. Each task may be mapped as being able to be completed by single or multiple different staff types. For example, one business may allow a general sales task to be performed by any staff type, they may require a store opening task to be performed by a manager or assistant manager, and may require a manager to perform a store closing task.

Rules

[0049] The system may specify a wide range of rules which constrain the permitted roster compositions. One category of rules will be legal rules that the business is legally obliged to comply with. The most significant sources of these legal constraints will be legislation. In Australia this legislation defines and governs the National Employment Standards, the relevant industry-specific award, or the business-specific enterprise agreement that the business is subject to. These legal rules specify constraints including, but not limited to a minimum or maximum number of hours that staff may work in a day, a minimum and maximum number of hours on a shift without a break, or the number of consecutive days that may be worked. However, within these constraints, the legal rules will allow the business flexibility to roster staff members on for different time periods as required by the business.

[0050] The legal rules may also specify staff pay rates (including penalty rates) for each staff type. Furthermore, although the present invention generally deals with staff types, rather than individual staff members, in some cases a business may have individual agreements with particular staff members, and there may be rules specific to a staff member which reflect that individual agreement.

[0051] The business itself may also have its own fixed business rules, which define the operational facts of the business. Fixed business rules in a retail example may include the number of stores, the size of each store, the opening hours, the location and turnover of each store.

[0052] Furthermore, the rules may also include workforce planning rules. Workforce planning rules may include employee productivity measures, or foot traffic counters within a store. These rules may also include associations between weather forecasts and sales data to map and forecast staff requirements. These allow the business to plan their workforce requirements. [0053] Furthermore, the rules may include flexible business rules. These include business-specific rules that can be chosen and varied by the business. For example, they may include the staff types that may perform a particular task - some tasks may require a particular level of seniority, or may require staff who are specially trained to perform the task. Figure 3 depicts examples of staff types being mapped to tasks.

[0054] Other flexible business rules may specify a minimum staff coverage required at a particular time of day. Other flexible business rules may specify preferred shift lengths for a staff type, or minimum and maximum hours per week for a particular staff type. The rules may also specify, for each staff type, minimum and maximum numbers of the staff type that may be included in the roster (generally to reflect the actual number of that particular staff type, employed by the business).

[0055] One significant constraint imposed by the business itself will be the coverage requirements for each task. Figure 4 is a table showing a weekly roster with defined tasks in 15 minute blocks, although naturally in different embodiments, the duration of the task block may change. For convenience, this table does not depict any concurrent tasks;

however, as previously mentioned, different tasks may be performed and therefore rostered concurrently. The coverage requirements are listed in each cell (showing the minimum number of staff required for that task). In practice, the coverage requirements are not likely to be met precisely, due to limitations caused by shift lengths and maintaining coverage while staff are on necessary breaks. However, although there will be instances where there will be more coverage than required, the coverage should never be less than the defined coverage.

[0056] Other business-imposed rules may include a requirement that a certain staff type receives a minimum number of hours, so that they can ensure that employees of that type receive a reasonable number of hours. A business may impose certain allowable times for training staff. Or alternatively, a business may require a certain staff type (e.g. manager) to work on specific days (e.g. weekends).

[0057] As another alternative, a business may have a preference for the flexibility offered by casual staff, compared to full-time or part-time staff. This flexibility may, in the business' view, offset the additional cost of casual staff. Therefore, rules may be implemented weighting the cost of preferred types of staff so that they are preferentially rostered over other types of staff.

[0058] By way of example, a general framework for the rules, within which the roster is designed or optimised is set out below:

1. Coverage requirements for the full roster cycle.

2. Award/EA (legal) requirements:

(a) Salary requirements (i.e. the total cost of a roster must not exceed the salary);

(b) Minimum and maximum work hours in a day and roster cycle;

(c) Minimum shift length;

(d) Days on requirements;

(e) Days off requirements;

(f) Maximum consecutive days;

(g) Break/rest requirements;

(h) Minimum rest between shifts;

(i) Shiftwork requirements;

(j) Span of hours and requirements; (k) 24-hour shift requirements; (1) Shiftwork requirements; (m) Split shift requirements; and/or (n) Training requirements.

3. Business rules:

(a) Business operating hours; (b) The minimum and maximum number of individuals for each role;

(c) Mapping of job classifications (in Award/EA) to roles;

(d) Task map for each role;

(e) Higher duty requirements;

(f) Floating tasks requirements;

(g) Blocked tasks requirements;

(h) Presence requirements for a managerial task;

(i) Presence requirements for a coverage task;

(j) Work hour limits;

(k) Work day limits;

(1) Mandatory work days (e.g. must work on . . . )

(m) Work day restrictions (e.g. if work on day . . . , must not work day . . . );

(n) Restrictions on when a shift must start and/or end;

(o) Maximum composition percentage of roles in the workforce;

(p) Role preference; and/or

(q) Special loadings.

Costs (could be defined in Award/EA or by the Business):

(a) Salaries;

(b) Base rates;

(c) Penalties rates (e.g. overtime, Saturday, Sunday, higher duties etc.);

(d) Allowances; and/or (e) Annual leave coverage.

[0059] Again, the above examples are illustrative only, and not exhaustive. In accordance with the present invention, a very wide range of business rules and industries may be specified, and these may be modified from time to time or to adapt them to specific businesses. Furthermore, the rules may allow for variations, such as by mutual agreement between staff and employers.

Modelling

[0060] In order to optimise the roster composition, the problem can be expressed as a number of variables, that can be varied within constraints defined by the rules, to determine their optimum values.

[0061] The Anonymous Roster Optimisation Problem (AROP) aims to find a cost- minimal collection of anonymous rosters that is both legal (with respect to the Award or Enterprise Agreement(EA)) and feasible (with respect to the business rules), and satisfies the given coverage requirement.

[0062] The set of inputs to AROP includes the rules specified discussed above. The AROP aims to minimise the total cost of the collection of rosters, where this is the sum of all the costs components (e.g. as listed above), such that the requirements of the other rules are satisfied.

[0063] Two approaches to modelling and solving the AROP are described below - a modular approach and a single-problem approach. The appropriate approach, for a particular roster, can be selected based on the complexity of the roster to be optimised and the available computing power.

[0064] The computer processing requirements for optimising the roster composition may represent a challenge for implementing the present invention, for some particularly complex rosters. The decomposition approach is likely to be more suitable where the roster to be modelled is complex (such as 30 or more staff) and there is (relatively) less available computing power. However, in different embodiments of the invention, different approaches may be taken to modelling and solving the AROP. Example 1

[0065] In this approach, the AROP is decomposed into smaller, digestible' subproblems or modules, which are solved in an iterative manner as described below. A summary of this modular approach is shown Figure 9.

Definitions

Shift: A continuous block of work, described by the start and end time of the work block, the start and end time of the break within the work block, and the task performed for each time period of the work block.

Shift skeleton: A shift with no specification of task(s) performed.

Work pattern: A collection of shifts that must be worked over a period of time.

Roster : The work pattern assigned to an employee.

Set of Rosters: The work pattern assigned to a group of employees.

Anonymous roster: A roster where the assigned employee is unnamed.

Coverage: The minimum number of employees required to be working on a particular task in each time period, for all tasks and time periods in the full roster cycle.

Role: The function assumed by an anonymous employee. For example, salaried (full time) store manager, full time assistant store manager, part time sales assistant etc. Generally, the role will be defined by job (e.g. Store Manager, Assistant Store Manager, etc) and the Work Type is defined as "full time", "part time" or "contract, etc. However, for illustrative purposes in the following examples, "role" has been defined as job and work type combination.

Individual: An anonymous employee.

Roster Composition: a set of anonymous rosters.

Stage 1: The Daily Shift Optimisation (DSO) problem [0066] Given a set of code-generated legally compliant shift skeletons (e.g. 3-hour shift, 5-hour shift, 6-hour shift with one 30-min break etc.), coverage requirements for a day, and relevant 'daily' requirements (e.g. minimum shift length), the DSO problem aims to find a set of shifts (to be used as input for Stage 2) that minimises the weighted sum of the total number of shifts used, plus coverage oversupply penalties.

[0067] The DSO problem is formulated as a mixed-integer linear program (the DSO Master Problem) and its solution checked for task-assignment feasibility (the Feasibility Subproblem) via a series of linear programming subproblems. If the Feasibility

Subproblem is infeasible, a 'corrective constraint' is added to the Master Problem and solved again. The Master Problem and Feasibility Subproblems are solved iteratively until no more 'corrections' are required.

[0068] The key set of variables for the DSO Master Problem is n - an integer variable describing the number of shift skeleton s commencing at time bucket t for role r. The daily coverage constraints are approximated in the Master Problem using these n variables and a set of variables representing the coverage oversupply. The approximation requires that enough 'people' are available to carry out the tasks, but does not assign the tasks (this is left to the Feasibility Subproblem). All relevant 'daily' requirements are modelled using n variables and additional auxiliary variables.

[0069] The Feasibility Subproblem consists of a set of softened coverage constraints and employee 'supply' constraints (derived from the solution of n variables) and the objective is to minimise the sum of the soft variables.

[0070] DSO problems for all days in the roster cycle is solved in Stage 1, and solutions to the n variables (shifts, not shift skeletons) are passed on as input to Stage 2.

Stage 2: Next, the Work Pattern Construction (WPC) problem

[0071] Given the set of shifts (Stage 1 output), and the rules determined above, the WPC problem aims to determine the set of cost-minimal anonymous rosters (roster composition) such that each and every shift is 'housed' in exactly one anonymous roster. The coverage requirement is not included in this problem since the requirement to assign all shifts implies satisfaction of the coverage requirement. [0072] The WPC problem is formulated as a mixed-integer linear program (the WPC Master Problem) and its solution is checked for feasibility using a program subroutine (the Feasibility Checker). The WPC Master Problem and the Feasibility Checker is run iteratively in a manner similar to the DSO problem in Stage 1.

[0073] The key set of variables for the Master Problem is x_is - a binary variable used to indicate that individual i is assigned shift s when Xj_S = 1; Xj_S = 0 otherwise.

[0074] Other auxiliary variables, along with these x variables, are used to model constraints related to the rules. Some of the constraints are approximations to the actual rules - this is the reason why the solution to the Master Problem is passed on the

Feasibility Checker. The objective function of the Master Problem is the approximation of the total cost of the work patterns.

[0075] The Feasibility Checker, a program subroutine, is an accurate representation of the rules and checks that the proposed solution complies with the rules. If the solution is infeasible, a set of 'corrective constraints' are appended to the Master Problem so that the same breach of the rules will not be repeated in the next iteration.

[0076] The output of the WPC problem, a set of anonymous rosters, is converted to a set of work patterns (by removing the individual associated with each anonymous roster) and passed on as input to Stage 3. The actual cost of the work patterns will be calculated in Stage 3.

Stage 3: the Roster Pattern Optimisation (RPO) problem.

[0077] Given the set of legally compliant work patterns (Stage 2 output), coverage requirements, and other rules (e.g. minimum and maximum number of individuals in each role, presence requirements etc.) the RPO problem aims to determine the cost-minimal set of anonymous rosters (roster composition) so that the coverage and business requirements are satisfied. The RPO is modelled as a mixed-integer linear program with key variables such as n_rp - an integer variable representing the number of individuals with role r using work pattern p. The actual cost of each work pattern is evaluated using the Costing Module (a program subroutine), and the compatibility/feasibility of pairing a work pattern with a role is evaluated using the Feasibility Checker. [0078] Other iterative processes may be used - for instance, a column generation procedure may use further subproblems in order to seek additional good' work patterns and append them to the RPO problem. The search for additional good' work patterns aims to reduce any 'gaps' caused as a result of decomposing the problem into smaller, digestible subproblems in Stages 1, 2 and 3.

[0079] Other alternatives may include using metaheuristics and shortest paths algorithms, that could be running in parallel, or used in replacement of methods described in Stages 1, 2, and 3.

Example 2

[0080] In another example, the AROP may be modelled and solved as one mixed- integer linear program.

[0081] This formulation models each anonymous employee explicitly, i.e. if we can have up to a maximum of 5 full time sales assistant, the formulation will have variables and constraints for employee 1, employee 2, . . . , employee 5.

[0082] For each anonymous employee i, each day of week d, and each week w, the key sets of decision variables are shift segment variables, days on and days off variables, and coverage variables.

[0083] Shift segment variables: A shift is defined by five segments, alternating between a work block and a break block, i.e. work-break-work-break-work. Attached to each work segment is a start time variable, and a set of duration variables (and a derived variable is also defined for the end time of the work segment). A duration variable is defined for each of the break segments. To ensure the validity of the shift constructed, a set of logical and accounting constraints are defined based on the shift segment variables. For example, if duration of work segment 1 equals 0, then durations of work segments 2 and 3 are equal to 0' is a logical constraint constructed to enforce the validity of a shift.

[0084] Days on and days off variables: One binary variable is defined to capture if employee i works on day d of week w, and another is defined for day off. These variables are 'logically' related to the work segment duration variables. For example, if the total work segment duration is greater than 0, then the days on variable is set to 1. [0085] A set of coverage variables is also defined for each time interval [α, β] each coverage task and each anonymous employee. These coverage variables are binary variables. It is necessary that time is discretised into time buckets so that the interval [α, β] can be defined in a finite manner. These coverage variables are related to work segment start and end time variables in a way that if the work segment overlaps with the interval [a, β] the coverage variable is set to 1.

[0086] In this embodiment, the roster will ultimately define work segments for each staff type, for each day within the roster time period. Each work segment may be specified by its start and end time, each segment being a continuous variable. Based on typical award or enterprise agreements, there may be up to three work segments, per staff member, per day.

[0087] Assuming there is more than one work segment on a given day, a number of paid or unpaid breaks may also be specified as continuous variables, in a similar manner to the work segments. Each break variable will have a start time corresponding to the end of the preceding work segment and an end time corresponding to the start of the subsequent work segment for the applicable staff type.

[0088] Figure 5 depicts a simple arrangement of three work segments and breaks in accordance with these defined variables in the model.

[0089] Coverage variables may be binary variables indexed by a starting time of a particular period, and may simply indicate (for each work segment) whether a particular task has been covered by that staff (in which case the cover variable is set to 1), or not (in which case the cover variable is set to 0). For each task, the sum of these coverage variables can be compared to the coverage requirement, thereby determining whether the coverage requirement has been met.

[0090] Finally, as described previously many other variables may be defined in a model utilised in accordance with this embodiment of the present invention. These arise out of the rules for the roster. In order to be considered and complied with during the optimisation of the roster, these rules should be expressed in a model or digital

representation of the rostering problem, as variables having associated constraints. Each of these rules will directly or indirectly affect the final roster composition. [0091] These constraints, which can be expressed digitally as equations or inequalities, may be based on an industry specific set of legal rules and as such may vary based on the clients applicable industry legal and employment rules. However, a skeleton structure may be provided that is applicable across a variety of industries.

[0092] It will be appreciated that any model has limitations. The model is therefore an approximate representation of the original problem (optimising a roster composition which ensures business compliance with legal rules governing employment of staff; interrogate, challenge, and validate assumptions around current coverage requirements, and then to provide a new optimised set of minimum coverage requirements in order to meet legal rules and business rules; provide insight and cost analysis of business decisions relating to labour deployment; and to minimise the cost of labour requirements). The accuracy of the final roster composition will be limited by how accurately the model reflects the real-world constraints imposed under the rules. The accuracy of the final roster composition will also be limited by the ability of the commercial solver engine to solve the equations provided for minimum cost. Alternatively, the solver engine may determine that it is impossible to impose all of the constraints, and identify which constraint is impossible to impose.

Cost Calculations

[0093] As will be appreciated, rostering on staff has an associated cost. This will be a minimum time rate (e.g. a cost per hour) as defined in accordance with the relevant award or an enterprise agreement used by a business, or this may be a higher rate as defined between an individual and the employer. Different (and higher) overtime rates also apply if a staff member has worked more than a specified number of hours in a row, or specified number of hours in a day as defined by the relevant award or enterprise agreement.

Furthermore, different (and higher) rates may also apply for specific types of work or at specific times. For example, under some enterprise agreements, weekend, holiday or late night working hours may require higher wages. Similarly, some tasks may be classified as a 'higher duty' task even if performed by a lower level staff type. These higher duty tasks will attract a higher rate of pay.

[0094] Furthermore, different (and higher) penalty rates and allowances (costs) may also be applied for specific task types as required by the employer and defined by the relevant award or an enterprise agreement. [0095] Figure 6 depicts a table showing the calculation of costs for a number of staff. As can be seen, there are a variety of different factors that may be taken into account when determining the cost incurred for a particular staff type on a particular day. In this example, the arrows show a part-time staff member who has worked outside of ordinary hours by 30 minutes, and so the penalty associated with this non-standard time is calculated and costed. In some cases, if excessive penalty rates are incurred, this may result in more junior staff being more expensive to utilize than more senior staff - for example, for scheduled hours on a weekend, a full-time senior employee may cost less than a junior employee employed on a casual basis and under an agreement that imposes penalty rates for weekend work.

[0096] Accordingly, shift worked by each staff type comes with an associated cost, depending on the value (i.e. duration) of time worked and the staff type, as well as any overtime or penalty rates that may be applicable. These rates and costs may be captured as part of the overall problem framework, as set out in paragraph [0058], heading 4 above. The total cost of the roster may be expressed as the sum of the cost associated with all of the variable values. Changing the value of these variables will result in corresponding changes to the overall cost of the roster - meaning that some feasible rosters (that comply with all of the business rules) will cost less than others.

[0097] In addition, where important business preferences are not incorporated into the engine and its cost function, 'fictitious costs' may be added to the engine's cost function that guide the engine to produce rosters that adhere to business preferences - even when these preferences do not have an immediate cost impact. For example, coverage is a binary variable. If a coverage rule is complied with, and the coverage variable is set to 1, then the cost may be $0. However, if the coverage variable is 0 (meaning that it has not been complied with), then a fictitious cost of $1,000,000 may be assigned to that variable. This facilitates the softening of constraints, which can be of benefit during the optimisation - it means that an initial set of variables (although not optimal) may still be considered

'feasible', and provide a starting point for the engine to begin varying the variable values and therefore more efficiently generate the optimal rostering solution.

Determine Desired Values of Variables

[0098] In light of the above, the present invention aims to minimise the cost of the roster, within the constraints of the rules. In one example (consistent with the second example provided above), the rostering problem can accordingly be formulated as an optimisation mixed-integer program. That is, to minimise subject to where x_t are the roster variables, c, is the cost of x_{, and a, and b, are the known coefficients specified by the roster structure and the business rules.

[0099] This type of problem is a complex and non-polynomial hard problem, which can be solved by known linear optimisation/evaluation engines. Specifically, IBM provides an optimisation/evaluation engine known as CPLEX, which will optimise a specific set of variables (in this case for minimum cost) within a defined set of constraints. Other commercial optimisation/evaluation engines are also available, including GUROBI and FICO XPRESS, and they could be used to solve the mixed-integer problem identified above, subject to appropriate formulation of the problem for those engines. Variables and the types of constraints presented in order to formulate the problem are discussed above.

[00100] Accordingly, a model of the rules, comprising the variables to be optimised and their associated constraints, may be provided to the engine. The typical approach of the engine is to solve the problem defined by the roster model by a combination of Operations Research algorithms and some appropriate heuristics without violating the governing constraints.

[00101] The output of the engine will ultimately be a set of values for the variables of the model, which can be interpreted to provide a roster coverage baseline with minimum total roster cost which also satisfies the governing constraints including but not limited to legal rules, fixed, and flexible business rules.

Roster Composition

[00102] The set of values for the variables can then be translated or interpreted, to have a meaning applicable to a roster composition - for example, a variable may correspond to a specific shift length and task assignment for a given staff type. As such, the output of the engine can be used to define a roster composition reflecting these optimised values. In particular, the desired values can be used to define shifts for each staff type. This defines an optimum composition of staff types, for the roster. [00103] Figure 7 depicts a roster composition generated in accordance with an embodiment of the present invention. In this embodiment, the roster composition does not allocate existing staff members to defined shifts. Rather, it is role specific, allocating staff types to coverage requirements for tasks, as shown in Figure 7. This enables the optimisation of the number and composition of staff types in order to meet minimum coverage requirements, in order to meet the minimum business and legal requirements. Figure 8 depicts an example of a coverage summary for this particular roster composition, with the number of allocated employee types listed on the y-axis. The roster composition can then be used to determine a baseline minimum roster cost for a roster which meets the coverage requirements and other business rules.

[00104] The minimum cost roster composition can be determined from the values of the work segments for each staff type, that are obtained from the optimisation/ evaluation engine. The output from the optimisation/evaluation engine is very raw, and to be meaningful must be interpreted and translated into an appropriate optimised roster composition. Accordingly, the desired values of the variables are interpreted to represent the allocation of staff types to tasks in the roster composition. By way of example, Figure 10 depicts how the output of the solving engine, a list of values, is first mapped to particular variables that correspond to a particular staff role and work pattern, and then those variables are used to generate an associated roster composition. For example, the 'Objective Value' in Figure 10 corresponds to the total roster cost. Furthermore, in the highlighted portion on the left side of Figure 10, the value T from the optimisation engine is mapped to the variable n[5, SO_RSOPT_7]. This can then be mapped to a particular role and work pattern - that is, there is one individual with role '5' (part time sales assistant) assigned to work pattern SO_RSOPT_7, as highlighted on the right side of Figure 10.

[00105] Once a roster composition has been generated, the business must assess how to legally migrate from the existing rosters and employee compositions within their organisation to the optimal roster composition should they choose to implement it.

[00106] In some embodiments of the present invention, the roster composition may subsequently be used to generate a roster (actual staff in actual shifts). In some cases, this roster may be generated automatically from the roster composition baseline, by

automatically assigning available staff members to shifts for their respective staff type as defined by the business rules. However, in other cases, the generation of the roster may be performed manually by a staff or business manager. For this purpose, the present invention may provide an interface that allows the manager to allocate specific staff members to specific shifts. This will allow the manager to make discretionary choices regarding the assignment of specific staff members to specific shifts. In such advanced embodiments, the present invention may be directly linked in to payroll tools and staff productivity tools, which may in turn feed back by providing weightings for use in the optimisation process.

[00107] Preferably, the system also determines an individual rule cost, which represents the cost of each defined rule. This will allow a business to understand the real cost of each rule, enabling the business to make informed decisions about the application of each rule, allowing greater control of their operating practices and the corresponding labor costs. For example, if a business has a rule requiring a particular task to be performed by a manager, the cost of this rule may be determined by comparing the minimum cost of a roster composition generated using this rule, to the minimum cost of a roster composition generated without this rule. Usually, these rules are interlinked so one change may not have a singular financial cost however the present invention will show an approximated scale of cost impact of altering or removing the rule in question.

[00108] Furthermore, different combinations of interlinked rules may also be compared. In particular, one advantage of this may be the ability to compare rules corresponding to an enterprise agreement with those corresponding to an Industry Award - an optimised roster composition may be determined for the business using a 'package' or 'set' of rules corresponding to each of these options (Industry Award or enterprise agreement), and the differences can then be determined (e.g. difference in baseline cost). This difference may be useful in ensuring that a business complies with its legal obligations with respect to enterprise agreements - in particular, enterprise agreements in Australia are required to be on terms no less favourable than industry awards. The output of the above comparison may inform a determination of whether a proposed enterprise agreement meets this requirement.

[00109] In a similar manner, different enterprise agreements may be compared to each other, to enable a user to make an informed assessment of the effect of the agreements. Alternatively, a comparison may be done between different sets of legal rules that would apply before and after a proposed change in legislation.

User Interface

[00110] Clearly, a wide range of user interface features may be required in accordance with the present invention. The interface features may be selected based on the particular type of user who is expected to use the software. For example a store manager will prefer a very different user interface to a specialist rostering consultant. The present invention may be configured for use by a variety of different potential users, and the user interface may be varied accordingly.

[00111] However, the user interface may, for some users, enable access across the full scope of the implementation of the present invention, from defining the problem to generating a specific roster. The level of detail accessible through the user interface will vary depending on the particular user - a line manager will preferably be provided with a very different experience to a system administrator. The examples depicted in Figures 1, 2 and 3 are of course designed to be utilised by an experienced person at a system

administrator level, rather than by a manager of a business.

[00112] However, a manager or HR manager will often be the best-placed person to provide the details of the business that must be reflected in the rules of the business. For example, an appropriate manager/HR manager at the business will know which Industry Award or enterprise agreement governs the business operations. Accordingly, the manager may be prompted to select the appropriate governing award/agreement, and may be presented with a choice from known awards/agreements. Furthermore, each of these known awards/agreements may have a predefined 'set' or 'package' of rules associated with it, which reflect all aspects of those awards agreements from permitted shift lengths through to pay scales and overtime provisions.

[00113] As another example, the system may prompt a user to define staff types, and allow them to define the tasks that must be performed (see the partial user interface provided in Figure 1). They may also be prompted to define further information such as the coverage requirements (Figure 2), staff/role mappings (Figure 3) and cost rules. [00114] Furthermore, the user interface may allow a user to view the optimised roster composition that is generated according to the present invention. Figures 7 and 8 depict two clear options for allowing a user to review a roster composition and understand how it reflects the coverage requirements.

[00115] The user interface may also be configured to allow the user to subsequently create a roster based on the optimum roster composition determined as described above. In one embodiment, the user interface may allow the user to define individual staff types, and assign them to work segments within the roster composition. The user may be restricted in only assigning staff of an appropriate staff type. Alternatively, suggested rosters, designating individual staff types, may be generated automatically and presented to the user.

[00116] Of course, the precise implementation of the user interface may be varied in different embodiments, and the invention is not limited to use with any particular user interface.

[00117] The present invention improves the process of roster generation. Firstly, it provides a scientific method of calculating and defining the minimum coverage

requirements for the purpose of workforce planning. Secondly, it automates the roster composition process, meaning that it is less labour intensive provided that the appropriate information is entered into the engine. Thirdly, it helps ensure that award and enterprise agreements are complied with. Fourthly, it can be used to identify the minimum roster cost, and associated staff type composition, in order to meet the legal rules and defined business rules, with the objective to reduce the overall cost of the business' roster, and allow the business to make informed choices about variance away from the minimum cost roster composition, thereby improving business control over labour cost and profitability.

[00118] In this specification where a document, act or item of knowledge is referred to or discussed, this reference or discussion is not an admission that the document, act or item of knowledge or any combination thereof was at the priority date, publicly available, known to the public, part of the common general knowledge; or known to be relevant to an attempt to solve any problem with which this specification is concerned.

Claims

CLAIMS The claims defining the invention are as follows:

1. A computer- implemented method for designing a roster for a business, the roster comprising a plurality of tasks and the business employing a plurality of staff types, the method comprising:

(b) creating a digital representation of the rules;

(c) deriving, from the digital representation, an optimal roster composition that provides a minimum labour cost to satisfy all the rules.

2. The computer- implemented method of claim 1, wherein the rules are determined by:

(d) prompting a user to specify a legal, fixed operational or flexible feature of their business; and

(e) receiving a user response specifying legal, fixed operational or flexible feature of their business, whereby the user response can be translated to one or more rules for the roster, the method thereby providing a systematised and scientific manner of capturing the rules to apply to the roster.

3. The computer- implemented method of claim 1 or 2, wherein the rules include at least one legal rule.

4. The computer-implemented method of claim 3, wherein the at least one legal rule comprises: a legislative requirement; an award requirement; or an enterprise agreement requirement.

5. The computer-implemented method of any preceding claim, wherein the rules include at least one fixed business rule.

6. The computer-implemented method of claim 5, wherein the at least one fixed business rule comprises: a number of stores or sites or facilities of the business; or required opening hours of a store or site or facility of the business.

7. The computer-implemented method of any preceding claim, further comprising at least one flexible business rule.

8. The computer-implemented method of claim 7, wherein the at least one flexible business rule comprises: permitted staff types associated with a task; or permitted shift lengths for a staff type.

9. The computer-implemented method of any preceding claim, wherein the digital representation includes

(i) one or more variables reflecting aspects of the roster composition, and a range of permitted values for each variable reflecting constraints associated with the rules;

(ii) a cost function associated with each variable, wherein the cost function depends on the corresponding staff type; and wherein the method further includes:

(f) determining desired values of the variables, within the constraints of the rules, to minimise the sum of the costs for all variables, whereby the desired values are used to select the staff composition which minimises the total cost of the roster and also satisfies the rules.

10. The computer-implemented method of any preceding claim, further:

(g) enabling a user to allocate staff members to the tasks in the roster composition.

11. The computer- implemented method of any preceding claim, wherein deriving the roster composition includes providing the digital representation to a solving engine.

12. The computer- implemented method of claim 11, wherein deriving the roster composition further includes solving the problem of determining a minimum roster cost with a modular approach, including providing digital representations of one or more sub- problems of the roster problem to the solving engine.

13. A method of calculating a cost associated with one or more rostering requirements of a business, the method including:

(h) calculating a first optimised cost for a roster composition with a first set of rules;

(i) calculating a second optimised cost for a roster composition with a second set of rules, the second set of rules having at least one differing rule which differs from the first set of rules, whereby the difference between the first optimised cost and second optimised cost indicates a cost of the at least one differing rule.

14. A computer-implemented method for checking the compliance of a roster composition with a plurality of roster rules, the roster composition allocating staff types to tasks over a roster time period, the method comprising:

(j) creating a digital representation of the rules, having a number of variables and having constraints for values of the variables;

(k) mapping the roster composition to the variables, to assign variable values to the variables; and

(1) checking whether the variable values are within the constraints.

15. A computer-implemented system for designing a roster, said system comprising one or more computers including: at least one processor; and at least one storage medium operatively coupled to said processor, said storage medium containing program instructions for execution by said processor, said program instructions causing said processor to execute the steps of the any one of claims 1 to 14.

16. A tangible computer-readable medium having computer-executable instructions stored thereon for directing a programmable device to perform the method of any one of claims 1 to 14.