If you translate the term „Data Science“ literally into German language it means „Datenwissenschaft“ – and this sounds very much like some stuff from the 1960s where computer filled halls with only a few MB.
„Datenwissenschaft“ sounds like nothing is left of the fancy analytical stuff, the AI, Deep Learning and all the other things you do having the sexiest job of the 21st century.
But rethinking it, this term meets more of what Data Science can do for us today than it seems. Yes, we need to do analytics. XGBoost, LTSM and tSNE can bring us insights into tons of data that contain information somewhere hidden behind rows and cols. Sometimes these gems are hidden in some easy spreadsheet which can be downloaded from the Datawarehouse which brave IT guys together with some even braver business analysts created in the early dawn of our young century. But most of the time a catalyst is needed to create value. This catalysts again are data. More data. From other sources.
This is not the ususal complaint of Data Scientists that 80 % of their work is data cleansing.
Matching data from different sources is the real gold
On the contrary. I believe that only the knowledgeable study of different data sources in their concrete context leads to the creation of business value from data. Understanding a business process related to a data generation process is an inherent need to each data analysis. But from someone who claims to be a „Data“ Scientists we need to expect more than understanding a single business process. Data Scientists should get a complete picture of their organization. With this knowledge they will be able to understand relations between different data sources.
Being knowledgeable about data
„Datenwissenschaft“ means to be knowledgeable about data processes. Understanding where data are generated, what errors and mistakes can be made generating data, knowing how to access, clean and match data. If we understand Data Science in this meaning of the term we will be able to open up our data silos and generate business value from it. Maybe by using some fancy analytics stuff. But maybe only by applying the right mix of data to our daily work.