Will the contributions of the data scientists to the solution of the scientific data be sufficient with the algorithms they have prepared?
The same algorithms as the other running programs will not always give the same results. Because the optimization methods and priorities are different. To get the result we want, we need to write the algorithm ourselves. So we have to be a programmer at the same time. We need to think like an engineer if we are going to do optimization. If we want quantum mechanical conformity we need to think like a physicist.
Turkish: Veri bilimciler hazır aldıkları algoritmalarla bilimsel verilerin çözümüne katkıları yeterli olabilecekler midir?
Diğer çalışan progamlarla aynı algoritmalar her zaman aynı sonuçları vermeyecektir. Çünkü optimazyon yöntemleri ve öncelikleri farklıdır. Buyüzden istediğimiz sonucu elde etmek için algoritmayı kendimiz yazmamız gerekir. Yani aynı zamanda da bir programcı olmamız gerekir. Optimazyon yapacaksakta bir mühendis gibi düşünmemiz gerekir. Eğer kuantum mekaniksel uyum istiyorsak bir fizikçi gibi düşünmemiz gerekir.
In my opinion no. The job of the data scientist is to turn data into knowledge. Developing an algorithm and throwing it out there will not do that. The data scientist needs to develop the computational solutions, and then apply them to extract knowledge from the data.
Also, there is some substantial gap between developing algorithms, and applying them on real-world data effectively. Data scientists are also expected to close that gap.
Unless a data scientist has a profound understanding in the field that the data are being extracted and presented for, his work may have just a partial success if any.
The best results are coming from a tight team cooperation of both subject matter specialists. On occasion, there are individuals who have a combined expertise in both areas, but it is rather an exception than a rule.
As others said already, it depends on the problem and data science is about solving a particular problem from the start to the end, which of course needs domain understanding. It is more than Algorithms as data acquisition and cleaning is also important.
IMO no as well. I started as an experimentalist. As my first student project, I received data files from a national laboratory, processed them in an algorithm I had designed and implemented under supervision, and send back the results to my supervisor. That seems to be the part the question asks about but there remains the jobs of explaining what the results mean and correlating the results with anyone else working independently on the same problem.
The same algorithms as the other running programs will not always give the same results. Because the optimization methods and priorities are different. To get the result we want, we need to write the algorithm ourselves. So we have to be a programmer at the same time. We need to think like an engineer if we are going to do optimization. If we want quantum mechanical conformity we need to think like a physicist.
Turkish: Diğer çalışan progamlarla aynı algoritmalar her zaman aynı sonuçları vermeyecektir. Çünkü optimazyon yöntemleri ve öncelikleri farklıdır. Bu yüzden istediğimiz sonucu elde etmek için algoritmayı kendimiz yazmamız gerekir. Yani aynı zamanda da bir programcı olmamız gerekir. Optimazyon yapacaksakta bir mühendis gibi düşünmemiz gerekir. Eğer kuantum mekaniksel uyum istiyorsak bir fizikçi gibi düşünmemiz gerekir.
- Badges
- Science topic