Worum es bei uns geht
This group is for those who believe in most of the following:
• data science is overhyped
• too many data science applications don't solve or even exacerbate today's most pressing problems. Examples can be found in advertising where the goal is to maximize revenue without necessarily taking into account whether the advertised product/service is sustainable (w.r.t. resource waste, customer health, working conditions, etc.). Another example is automated news generation because people don't need more news nowadays, but good quality news, enriched with in-depth discussions.
• there is too much focus on less important aspects of data science, like the choice of the machine learning algorithm, rather, one should focus on the more important aspects such as: data management (obtaining, validating, cleaning data, etc.), robust data-consuming processes (robust under missing & invalid data, the timing of other processes, etc.), data protection, choice of model and validation (instead of just blindly feeding more data to a model -> bias/variance tradeoff), company culture, etc.
• at some point data science exhibits diminishing returns, not least because the cost for maintaining and extending associated data, software, and documentation, becomes more costly than the gains from it
• to make data science more sustainable, practitioners must exchange best practices, especially when it comes to data management
Thus, if you want to calm the data science hype, hear and talk about truly useful applications, focus on its most important aspects, learn and convey its limitations, and exchange best practices, then this group is for you! In our Meetups we will present concrete data science use cases and discuss the above aspects for them.