How can your R&D team use data-driven approaches to problem solving?

In my experience inputs in R&D projects must be data driven. Sometimes R&D is colecting and analysing data, but it is more frequently Q dpt, that does that job. We should analyse problems we had in Manufaturing process (scrap, rework or FPY) and later by end customer (top faults, failure rate, reliabilty, user experience/customer satisfaction,...). With that inputs new concept and design must be improved for better quality.

From my experience, R&D don’t have much to do with troubleshooting etc in a factory. Once the product has been launched, the tech team on site takes over the management of the product, processes, testing and any deviation and root cause analysis. What does help with troubleshooting is a risk assessment done at the beginning of the project @ stability batches or scale up batches - to ID and mitigate any critical risks. Bad situation if you are trying to rectify issues on the run.

One way to clearly define the problem is to understand the real needs of the end user. Innovation in itself can be misleading if one does not clearly identify the needs of their targeted end user.

Toda inovação bem sucedida parte do princípio de resolver uma dor (problema) do mercado. Dor essa que muitas vezes o usuário nem sabe que tem, até que lhe apresentem uma nova solução. Então, partir de um problema é sempre uma ótima maneira de mapear soluções efetivas de grande sucesso e aceitação.

El equipo de investigación y desarrollo debería aprovechar enfoques basados en datos mediante el análisis riguroso de información relevante, implementación de modelos predictivos y toma de decisiones respaldada por evidencia. Esto nos permite desarrollar soluciones efectivas y optimizar constantemente nuestros procesos.

This is part of the problem definition. Are we going to develop a better solution than existing one, or are we going to develop a NEW solution for an unsolved problem?

Quantifying the as-is state creates a benchmark to evaluate progress. I gather relevant metrics using simulations, tests, surveys and other data sources tied to project goals. For a website redesign the baseline KPIs may include page load times, conversion rates, and satisfaction scores. For a robotic process automation implementation I would capture completion rates, cycle times and quality error rates. The measurements depend on the priorities - Is it speed? Accuracy? Adoption? Comparing solution iterations against the baseline highlights leading indicators of improvement towards targets. Periodic measurement also uncovers early performance regressions.

When revamping a manufacturing process in a product development cycle, we used data to establish a baseline for efficiency and quality. Through IoT devices and data logging, we tracked production metrics like cycle times, error rates, and material wastage. This provided a clear picture of the current process performance, enabling us to set measurable targets for improvement.

In the rapidly evolving landscape of electronic tape manufacturing, staying ahead requires a proactive and strategic approach to problem-solving. J World Electronic recognizes the pivotal role that data-driven approaches play in empowering the Research and Development (R&D) team to navigate challenges and drive innovation. Here's a closer look at how My R&D team can leverage data-driven methodologies to enhance problem-solving and propel the company forward: 1.Data Collection and Analysis 2.Predictive Modelling 3.Quality Control Optimization's 4.Innovation through customer Insights 5.Supply Chain Optimization 6.Continuous Improvement Culture

Medición si fuese algo netamente cuantitativo, pero también hay enfoques mixtos o cualitativos. La idea creo sería conocer hasta donde llega el estado actual de conocimiento y sobre ello seguir avanzando.

Mantente abierto a las posibilidades y reinterpretar los fenomenos. Si bien, muchas veces las variables principales tienen la respuesta; en otras ocaciones estas observaciones no describen las anomalias del sistema. Cuando te encuentres en un camino sin salida, observa. Hay condiciones satelites que no se estudian y pueden ser la pieza que estas buscando. Como en el caso de GM, el problema del motor puede encontrarse al estudiar la compra de helado de vainilla.

In a pharmaceutical R&D project, we employed data analytics to sift through vast molecular datasets. Using machine learning algorithms, we identified patterns in chemical structures correlated with desired biological effects. This data-driven approach helped us narrow down potential compounds for drug development, significantly reducing the time and resources needed for experimentation.

The pivotal step of data analysis necessitates meticulous documentation of the experimentation process. This involves a clear articulation of the utilized materials and the explained procedures. Once data is gathered, the next critical phase involves interpretation through analytical reasoning. This robust interpretation forms the foundation for subsequent steps, allowing contributions to the understanding of the subject. The essence lies in recognizing that small details, that collectively contribute to the bigger picture.

Analysis of data is an impactful parameter for the decision making/ or deciding the path of solution. To visualise the data ( experimental/databases) using different perspectives/view points - based on statistical analysis/clustering/to find out a pattern/ establishing a co-relation among different parameters will be helpful to identify the biological dynamics. It would be helpful to drive the solution in a context dependent manner.

Machine learning and modeling are two of the many tools that allow for data analysis to be optimized. These processes allow for rigorous analysis, free from bias. This is absolutely essential for data integrity. A danger of relying upon machine learning and emerging technologies is that it could lead to over reliance. We should employ all available tools at our disposal while maintaining our curiosity to better understand the ‘how’ and ‘why’ aspects of our research endeavors.

A cette étape il peut être intéressant de réaliser une première preuve de principe avant de se lancer dans un long développement qui peut être coûteux.

During the development of a new algorithm for image processing, continuous data analysis was crucial. We regularly tested iterations of the algorithm using real-time data feeds and performance metrics. By comparing results against predefined benchmarks and user-defined parameters, we fine-tuned the algorithm for higher accuracy and efficiency, ensuring it met the specified requirements before deployment.

A data-driven approach in R&D for involves leveraging data to enhance accuracy, optimize algorithms, adapt to different environments, and continuously improve performance based on real-world feedback.

No siempre es para mejorar una solución, es para incluso denotar que el avance es limitado o simplemente hay estancamiento. Hay diferentes niveles de pesquisa, no todo conllevan una mejora de solución, tomar nota.

Also here the completeness and availability of the understanding gained by any data analysis is crucial for further use. The more computationally accessible these insights are, the more automation is possible down the road.

Focussing on a given problem should not bar the mind from thinking out of the box and asking 'stupid questions'. There are no stupid questions. Instead, there are stupid answers like for example answering the question why something is done in a certain way by stating that this always has been done this certain way. Last not least, good communication and documentation is everything!

Data driven approaches are the most reliable and efficient way of solving problems in my opinion. It is important to have a holistic view of the problem to better construct the hypotheses to be tested, in order to avoid any bias that takes the team away from solving the problem.

Encourage collaboration between R&D and other departments, combination expertise from different area can provide a more comprehensive understanding of the problem and potential solutions.

By incorporating data-driven methodologies at each stage, our team was able to not only solve problems more effectively but also make informed decisions and optimize solutions for better outcomes.

Ensure ethical considerations and data privacy principles are prioritized throughout the data-driven approach. Maintain transparency and comply with relevant regulations to build trust with stakeholders.