Read on Twitter
I'm auditing a course on network analysis (in principle, fascinating and powerful with lots of applications), and the first assignment sample data is... employee relationships, where -100 = enemies and +100 = friends.
I just. What. No.
I just. What. No.
So I'm in a mood and decided to write down my own questions:
Why would an employer have data on employees' relationships?
Assuming such a thing is possible, is it ethical?
Where would this data come from? (2/5)
Why would an employer have data on employees' relationships?
Assuming such a thing is possible, is it ethical?
Where would this data come from? (2/5)
How are relationship values, from -100 = enemies to +100 = best friends, defined?
How are such values measured?
Have you thought that perhaps human relationships can't be quantified in such a way? (3/5)
How are such values measured?
Have you thought that perhaps human relationships can't be quantified in such a way? (3/5)
Given all of the above, what meaning would a Pearson correlation coefficient between these "relationship scores" and "movie preferences" even have?
Shouldn't data scientists be trained to ask these questions of their data? (4/5)
Shouldn't data scientists be trained to ask these questions of their data? (4/5)
I am not hopeful these questions will be adequately answered. The instructors will fail their first assignment. (end thread)
