Here's how you can effectively analyze and define problems as an Operations Research professional.

Talk to many people engaged in "the problem". What you will find is that each person experiences the "real problem", but then filters that through their own perceptions to describe "their problem". If you only talk to the Big Boss, you will only get the Big Boss's view of the problem. And that view may be distorted. Additionally, the individual problems that people describe are often tainted by minor organizational or interpersonal issues. Taking time to understand the "real problem" can make the issue simpler AND benefit a wider number of people. So talk to the folks on the production floor. Talk to the folks in the break room. Talk to the boss's secretary, the CFO, and the intern.

Je suis tout à fait d'accord, mais j'aimerais ajouter un commentaire : Pour commencer, il est essentiel d'être exhaustif en élargissant d'abord le périmètre afin de comprendre les interdépendances entre les différents problèmes au sein de la chaîne. Sinon, il y a un risque de résoudre un problème spécifique sans synchronisation avec d'autres problèmes interconnectés, ce qui pourrait soit accroître la difficulté globale, soit ne pas apporter d'amélioration significative

Lo primordial en este caso es tener una visión clara del problema, pero no sólo a nivel superficial, ya que eso podría ser sólo la punto del iceberg, sino a nivel más profundo: por ejemplo, un error en un dato, igual no es problema del dato de origen, sino de una transformación errónea. Tenemos que desenmarañar el problema hasta dar con el origen real y así poder evaluarlo bien y evaluar el alcance y consecuencias. Esto sólo se consigue conociendo los procesos que podrían haber dado lugar al error, por tanto es altamente recomendable tener buena documentación sobre cada proceso que se desarrolle.

Start by conducting a thorough needs assessment to gather relevant data from multiple sources. Real-world challenges often include incomplete data and conflicting stakeholder interests. Use root cause analysis to dig deeper into underlying issues rather than surface symptoms. For example, if a production delay occurs, investigate whether it's due to supplier inefficiency, equipment failure, or workforce shortages. Employ decision trees to visualize potential scenarios and constraints. By accurately defining the problem, you set a solid foundation for developing robust models and strategies that address the core issues, ensuring your solutions are effective and sustainable.

Challenges often arise, such as incomplete datasets or conflicting sources. Employ data cleansing techniques to rectify these issues and ensure consistency. Use SQL for efficient data extraction and transformation. In cases of missing data, apply imputation methods to fill gaps. Leverage tools like Python or R for statistical analysis, ensuring robust data interpretation. Combine structured and unstructured data to gain comprehensive insights. Address data privacy concerns by adhering to relevant regulations. Accurate data collection is crucial for reliable models and effective problem-solving.

Start with clearly defining objectives and constraints—whether it's minimizing costs or maximizing throughput. Real-world data is often messy and incomplete, posing challenges for accuracy. Tools like mixed-integer programming can handle complex, discrete decisions, while dynamic programming is useful for sequential decision-making under uncertainty. When dealing with stochastic variables, Monte Carlo simulations provide insights into variability and risk. Ensure your model is scalable to adapt to changes in data or objectives. By focusing on precise, relevant variables and employing the right techniques, you can derive actionable strategies that enhance efficiency and effectiveness.

A guiding Northstar is always important to assess solutions. Hardened OR professionals usually have a knack for identifying what the yard stick needs to be to select the best solution out of many approaches available For new grads, building that sense is usually the hardest. One thing that can help is to gain expertise in your chosen industry. For example in consumer electronics , the complexity of the solution chosen (and the time available to analyse and optimize) both scale with the complexity of the planning at hand. For many logistics problems, speed of execution is more important which lends to more heuristic based approaches for routing, while with strategic planning - a solver/simulation based approach is more applicable