How can you determine the scope of a data analytics project in OR?

In my experience as a supply chain consultant, to effectively comprehend an Operational Research problem, adhere to these best practices: 1. Clearly articulate the objectives and goals of the project. Understand the current situation and envision the desired outcome to establish a roadmap for analysis. 2. Uncover the main challenges and opportunities inherent in the operational context. This provides a comprehensive understanding of the factors influencing decision-making. 3. Engage Stakeholders. By incorporating these practices, you establish a robust foundation for understanding the Operational Research problem, fostering effective communication, and ultimately ensuring the success of your data analytics project.

In my role as a data analyst at an investment firm, I led a project to enhance portfolio performance by addressing underperforming assets and unclear market trends. The goal was to provide investment managers with data-driven insights for strategic decision-making, aiming to improve returns for our clients. Focusing on portfolio analysis and market trend prediction, the project aligned with the firm's objectives, maintaining a clear scope to avoid scope creep and directly contribute to the firm's performance goals.

Here’s some steps to consider which are important in understanding and formulating a sequential decision analytics model: 1. What is the problem description in plain English? 2. What metrics are we trying to impact? 3. What are the decisions? 4. What are the sources of uncertainty?

The continuous expansion of client requirements and project deliverables, is a significant challenge faced by project managers. It poses a significant obstacle as it results in an enlargement of the project scope, often leading to extensions in both the project timeline and budget. The underlying causes include inadequate initial planning, misconceptions client requirements, and non-compliance with the scope plan. The absence of a clear system to handle and control change requests exacerbates the issue. Project managers must anticipate and prepare for the inevitability of scope creep. A detailed understanding of client requirements is crucial, and acquiring skills in business analysis may prove beneficial in managing troubles effectively.

In my experience, the initial phase of scoping a data analytics project in operations research (OR) involves a comprehensive understanding of the problem at hand. It's crucial to delve into the intricacies by asking key questions, such as assessing the current situation and envisioning the desired outcome. Identifying the main challenges and opportunities, pinpointing decision makers and end users, and understanding their needs and expectations are pivotal steps. Moreover, aligning the project with organizational goals and strategies is essential for ensuring its relevance and impact. By meticulously clarifying the problem statement and project purpose, you not only define scope boundaries but also safeguard against potential scope creep.

Gold plating involves the intentional addition of features or functionalities not initially specified in the project scope statement. Project teams often implement gold plating at no extra cost to please the client. While it can be appreciated by some customers, there are instances where adding unrequested features may be undesirable, potentially constituting an unauthorized change in scope. Common in programming, gold plating is often done to showcase team skills or to enhance client satisfaction. Both scope creep and gold plating can impact a project's scope and ability to achieve objectives. Gold plating is more easily avoidable with careful project management.

In my experience, the subsequent phase of scoping a data analytics project in operations research (OR) centers around a meticulous review of the data landscape. It's essential to evaluate the existing data and the necessary data to be collected, analyzed, and modeled. This assessment includes gauging data availability, quality, relevance, and suitability for the project. Additionally, considerations extend to data sources, formats, sizes, and security aspects. Key questions arise, such as identifying the data on hand and the data required, determining access, storage, and management protocols, and devising strategies for data cleaning, transformation, and integration.

Most of the problems in real life are decision making under uncertainty such how many products to order, what products to promote? Yet I see a lot of trying to use deterministic optimization methods or combination of deterministic optimization and machine learning (predict then optimize) to solve problems. Please stop! You need to use a sequential decision analytics framework which may or may not include deterministic optimization methods.

Delve into the interpretability of the methods, ensuring that the insights derived can be clearly understood and communicated to stakeholders. Strive for a balance between sophisticated techniques and a comprehensible presentation of results. In essence, the appropriateness of selected methods is foundational to the success of a data analytics project. It involves a nuanced understanding of the problem, data, and objectives, ensuring that the chosen methods are not only technically robust but also strategically aligned with the unique context of the Operations Research project. This careful consideration sets the stage for a robust analytical approach and increases the likelihood of achieving meaningful and actionable insights.

In my experience, the pivotal third step in scoping a data analytics project in operations research (OR) revolves around method selection for data analysis and modeling. This involves a thoughtful consideration of techniques, tools, and frameworks that align with the specific problem, data characteristics, and project objectives. Evaluating the feasibility, complexity, and validity of chosen methods is essential. Critical questions emerge, such as understanding the assumptions and limitations, devising strategies for testing and validation, measuring and comparing method performance, and establishing clear documentation and communication protocols.

In my experience, the crucial fourth step in scoping a data analytics project in operations research (OR) entails meticulous resource planning. This involves estimating the time, cost, and quality parameters of the project. Identifying the necessary skills, roles, and responsibilities within the project team is equally vital. Key questions arise, including the duration of the project, milestones, and deadlines, cost estimation, budget considerations, and funding sources. Ensuring the quality and accuracy of project outputs and outcomes is a priority, along with clarifying the involvement, tasks, and expectations of each team member.

Operations Research (OR) projects are optimization problems. So, they fall under Prescriptive Analytics. The recommendation needs to be based on multiple constraints and objectives. Final results would be a set of feasible prescriptions (generated in a combination of textual narratives, tables and charts) that helps the clients to perform better in the future, based on the past and present data. The value is measured when the suggestions fit the customers’ budget for improvement and after implementation (of the delivered model) they get their return of investment. Simple.

In my experience, what I learnt is that a recommendation system with Operations Research algorithm would be narrated with the language and vocabulary of the specific domain or business (and not mathematical or statistical terms). One can leverage the power of Generative AI, to generate such reports. Kind of human like suggestions for optimized outcomes.