The story behind max likelihood .... fully explained! Sigmoid Video : • The Sigmoid : Data Sci... Logistic Regression Video : • Logistic Regression - ... My Patreon : www.patreon.co...
I swear to god, when I was searching for maximum likelihood estimation, I was hoping you would make a video on this, scrolled down a bit and bingo!! I see Rithvik Math. My happiness knows no bounds. Thank you for existing.
Thank you so much for preferring to actually help people understand concepts rather than throwing fancy multi-step jargon that gets people frustrated at math topics. I would do anything I can to help keep the channel alive and I will recommend your channel to all my friends and colleagues. Thanks again and keep the excellent work.
As always, great videos! I mainly use them as supplement to masters level courses. What I love are the parts where you cover the intuitions about the formulas, it makes them so much more understandable. Keep up the good work!
Thank you very much. Been struggling to get the intuition of MLE and you helped me to understand it. Would be awesome if you could do a video on GMM estimation. Thank you!!!
Thank you for the video! I wanted to say they are great for revision, but I usually learn a lot more than I did in school on the topic... so thank you!
excellent explanation, very good as an introduction. Can you make something which delves into maths even more? Explanation of the last part it necessary and would be fun to understand. :)
In English most people use the terms Probability and Likelihood interchangeably - I can't help but correct this when I hear it. One nuance is that the Maximum Likelihood result is often very small, meaning the parameter value isn't very likely, it is just more likely than the alternatives. Ranked Choice Voting is designed to promote the Most Likable Choice past the 50% threshold. Great video as always; I hope you become a stats and data science professor!
Great & helpful video ! From 12:40 says : "This's the probability of seeing all the real world outcomes that i actually see in my data". I think its better to replace "real world" with "predicted" or other synonym since the former kind of mislead the viewers to think those "outcomes" are recording of happened event which actually not, they stem from the model prediction, it's the "data"(y:nx1) that recorded the real result. and from 20:05 that's the correct way to express it.
Your videos are absolutely helpful!! You're soooo damn good teacher and really good at delivering the complicated lessons into the easier way to digest. I hope I can pass this semester with your videos bc without your video it would be even worse! 😂 THANK A MILLION ❤️
exactly my thoughts. From what I read from other sources, Likelihood is given by L(parameters or distribution|observed data). Perhaps @ritvikmath can explain better
What should be the probability P(y_i| x_i, beta) where actual y_i is reject(= 0). If P(y_i| x_i, beta) is close to 0 or 0. Max Likelihood estimator will not be max even though beta parameters are fine tuned.
Question: In 8:03, you're introducing the conceptions of the likelihood, which from my understanding is the probability of real observation emerging given the y; but in the formula, it looks like the posterior probability, not likelihood, is there something missing?
I am confused. You are saying the probability and likelihood are the same according to what you have written by your equations. For likelihood are you not trying to say given the acceptance into med school the likelihood of having these values for gpa, mcat score etc…For instance if probability is P(y|x) then likelihood is L(x|y)? You have these two being equal.
while it's true that derivative=0 could mean min or max, we can distinguish since a min has a decreasing gradient on the left and increasing gradient on the right. a max is the opposite. hope that helps!