People often ask me what ‘data science’ actually is. At Amsterdam Data Collective (ADC), data science has a broad meaning: it is everything that has to do with the process of retrieving useful information from data, as well as editing and presenting that data. Some experts in data science will put the emphasis on the scientific basis, but we prefer a more pragmatic approach. By doing so, other elements become more important, such as the reliability of the method, the usability, and the predictions made by data analysis.
Data science or data art?
The ‘science’ in data science stands for the smart, efficient and reliable way in which data is processed into valuable information. However, this process remains challenging. Therefore, it is necessary to know its limitations. Data analysis does not often provide a solution to a problem right away. It may generate results, but eventually human intervention is still needed. At the end of the day, someone still has to interpret those results and decide what to do next. However, due to artificial intelligence, this seems to be changing as well. Think of robots and self-driving cars: the only thing humans still have to determine are the limitations of the programme, within which it can take decisions and learn on its own.
What use does it have?
The logical follow-up question is: what can you do with data science? Well, in short: answering tough questions by means of, for example, predictions, so that organisations are able to take more informed decisions. The process of decision-making used to be far more ‘experience based’. Someone with thirty years of experience was the go-to guy, because he knew how everything worked and was therefore able to take the decisions.
At the time, it was nearly impossible to determine whether the decision made was in fact the best decision that could be made. The availability and analysis capacity of much more data now enable us to perform this check, which not only helps to accelerate reproducible decision-making processes, but limits the risks as well. The right data-driven decisions produce better results and fewer errors. Moreover, they ensure that decision makers gain more time to use their knowledge, experience and creativity for other issues that have not been solved yet and still lack available data.
More than just numbers
As many data-driven questions show similarities, we continue to learn more about which algorithms should be used to solve a certain problem and how we should interpret the results. There are quite some professionals nowadays who know how to work with a pre-programmed algorithm and are therefore able to make a prediction, but the reliability and the novelty of that prediction depend in large part on the competence of the data scientist. We insist that our consultants look further than the numbers and data, because in the end, it’s all about finding the right solutions for management related issues and the applicability of those solutions within the organisation.
Our data scientists often work together with the client’s ‘in-house quants’. This is often favoured over fully outsourcing a project, as it allows you to benefit from their involvement and the knowledge available within the company itself. Moreover, once they have implemented the tool, companies tend to embrace it and eventually learn to make optimal use of it. Because that is our ultimate goal: providing solutions that actually make a difference.
Making an impact
At the end of the day, I would like our work to provide insights that are valuable for the client. For example, we worked with our entire team on a pro bono case for Stichting Hartekind, a foundation that helps to finance research on heart defects among children. We investigated if it is possible to determine whether children who have undergone cardiac surgery at an early age run a smaller or a larger risk of heart failure when they are adults.
When we could finally make a prediction, we were more than thrilled, because of the benefit it can bring to the foundation and the children they support.