1. Fixed my Raspberry Pi Zero W. Maybe some Pi project next! 2. Finished the first subcourse in Advanced Machine Learning 3. Spent a good amount of time 'clouding' the Clickbait project since I'm kind of tired of running stuff locally (Not only my laptop keeps overheating, but it also only charges when it's off. Unforeseen consequences of this are forced breaks every couple of hours, but less procrastination since even watching videos is too much)
I think I'll take a break from ML by taking the Advanced Machine Learning specialization course by HSE on Coursera. l o l
Basically the training is very unstable. Several reason: small data set -> overtrains very quickly metrics are sometimes somewhat misleading due to the noise in the dataset Also the CB signal itself seems to be very weak While we did achieve decent f1 scores, after closer inspection I think part of it was due to bias, thus signal is actually weaker
The best model achieved 0.69 AUC and 0.71 F1 score! We are getting there
YESYSEYYSEYEYSEYESYYSEYSE!!!! CLICKBAIT EXISTS!!! AOC 0.59 ACHIEVED WITH SOME RANDOM MODEL SUPER HAPPY!!
