Blogs

The 7 drivers of effective decision optimization

Decision Optimization Offering Manager, IBM

Many organizations face challenges when developing realistic plans and schedules that balance goals and constraints across diverse business functions, such as human resource planning, supply chain management and production sourcing. 

In this regard, decision optimization—also known as mathematical optimization or prescriptive analytics—can deliver huge value to planners and decision-makers. To drive coordinated planning across diverse business functions, the most efficient approach is to use common decision optimization tools that address business and process specifics.

An enterprise-wide decision optimization infrastructure should satisfy the following key requirements: 

  • Customizability: The common infrastructure should be highly customizable to an organization’s business and IT requirements, providing a foundation of common planning and optimization functionality. The platform should offer flexible and tailorable business process support, integrated decision recommendations and optimization, collaborative planning, scenario sharing and analysis, sensitivity analysis and explanations, IT integration architecture and scalable deployment.
  • Multiple objectives and goal programming: In the real world, business objectives such as minimizing costs, improving customer service, increasing revenue and profitability often conflict. The decision optimization environment should help balance these conflicting goals, allowing users to change the weights associated with each goal or set individual goals or limits.
  • Controlled relaxation of constraints: The common decision optimization environment should automatically relax over-constrained problems during run time. This would help ensure that a solution is always found and presented along with information about relaxed preferences or constraints. The optimized solution, with its recommended plan or schedule and attendant metrics, should be easily explorable, helping users to understand the optimization model’s dynamics and identify solution scenarios.
  • Scenario management: To support scenario comparison and reporting, the common decision-support infrastructure should provide complete storage of each scenario, including the input data, scenario parameters, goals, decision variable values and solution metrics. Data tables should include local storage for external data so they can be modified without changing data sources. Furthermore, scenario metrics stored in a database can be invaluable for tracking operational and financial performance over time. Applications running in that infrastructure should provide easy control of business goals, as well as both tables and charts for KPI and solution analysis.
  • Collaborative planning: The common decision-optimization infrastructure should support large-scale applications with remote planners and distributed planning processes. Planners should be able to share proposed plans with reviewers through the shared scenario repository, get feedback on a schedule through Excel or email, or facilitate joint data entry and validation. A scenario repository should ensure data safety during scenario editing through locking.
  • http://www.ibmbigdatahub.com/sites/default/files/decisionoptimization_embed.jpgEfficient application development: The decision optimization infrastructure should save development time by natively supporting all stakeholders, from IT and Business to Engineering and Operations Research. It should automatically generate a complete decision-support application from configuration descriptions that they provide. It should help business users take part in the iterative application configuration and development. Mathematical models should be developed in lockstep with data access, GUI configuration, server setup and application integration.
  • Uncertainty-aware decision support: The common decision optimization infrastructure should provide a dedicated tool to create plans that anticipate the future and hedge against uncertainty in future scenarios, to reduce reactive changes, increase margins and reduce response time. It should enable business users to compare and evaluate alternative plans based on KPIs and risk-reward trade-off across future scenarios.

To explore a solution that addresses all of these requirements and can be deployed on cloud-based environments such as IBM SoftLayer, try the IBM Decision Optimization Center, which supports the offloading of optimization jobs to IBM Decision Optimization on Cloud through specialized custom tasks.

Also: