Different ML goals lead to different evaluations of ML model effectiveness
Some thoughts about different ML projects
Interesting case happened to me. I was preparing for the, let's say, call for the ML talk in the Gambling industry. To be honest, I've never played casino, so I was a bit unsure about my skills.
Right now, I am working on ML/AI recommendations for subscription-based apps, so my goal is to help my users (apps' owners) increase their revenue using our recommendations. The goal could also be to increase the number of subscriptions or active users.
I started searching for the goals that ML should achieve in the Gambling area of business. It seems the aim is to keep players constantly engaged with the game by analyzing their behavior. However, here is the interesting thing!!! We need to use stimuli all the time like rewards, risks, and responsible gaming behavior (who knew it exists?) to set the algorithms in the right way.
In my work, we don't have such stimulation attributes for the algorithms. That's why it leads us to the interesting summary.
In my current work, I am mostly using linear regression and association rule mining (such terms!) for data exploration, while for the gambling industry, you need to use decision trees(mostly). Moreover, I always approach ethical issues with great attention, so here I would prioritize responsible gaming behavior or those rewards, risks, etc. 🎲 Can't wait to see what to expect😄
