This talk is a continuation of the previous talk (part 1), which introduced recommender systems. watch it here: ruclips.net/video/Ams4sEn50cw/видео.html&ab_channel=DataScienceDojo
Thanks for architecture. A lot of tutorials talk about model itself!. But I couldn't quiet get recommended items are in cache or you get candidates and rank items in real-time? Thanks in advance
Excellent resource! Really enjoyed the depth covered in this hour-long video. On the candidate generation model: how is a simple ML model trained on users, i.e., what are the inputs and output(s) for that simple ML model?
Hello Hasnain, a simple machine learning model for candidate generation is typically trained on a dataset of user interactions with items, such as clicks, views, or purchases. The inputs to the model would be features extracted from these interactions, such as the user ID, the item ID, the timestamp of the interaction, and any other relevant contextual information. The output of the model would be a probability score for each item, indicating the likelihood that the user would be interested in that item. This probability score can then be used to rank the items and generate a list of recommended candidates for the user. Hope this helps!
Absolutely, it can be really frustrating when the recommendations don't seem to align with our tastes. Netflix uses complex algorithms based on a variety of metrics, including viewing history, user ratings, and even time of day you watch. Sometimes, though, it feels like these don't capture our preferences accurately. Have you tried tweaking your profile or rating more shows and movies? It might help refine what's suggested to you. Also, it's interesting to think about how different users experience these systems differently.
This talk is a continuation of the previous talk (part 1), which introduced recommender systems. watch it here: ruclips.net/video/Ams4sEn50cw/видео.html&ab_channel=DataScienceDojo
Thanks for architecture. A lot of tutorials talk about model itself!. But I couldn't quiet get recommended items are in cache or you get candidates and rank items in real-time? Thanks in advance
Excellent resource! Really enjoyed the depth covered in this hour-long video.
On the candidate generation model: how is a simple ML model trained on users, i.e., what are the inputs and output(s) for that simple ML model?
Hello Hasnain, a simple machine learning model for candidate generation is typically trained on a dataset of user interactions with items, such as clicks, views, or purchases. The inputs to the model would be features extracted from these interactions, such as the user ID, the item ID, the timestamp of the interaction, and any other relevant contextual information. The output of the model would be a probability score for each item, indicating the likelihood that the user would be interested in that item. This probability score can then be used to rank the items and generate a list of recommended candidates for the user. Hope this helps!
For further tutorials on advanced machine learning, check out this exclusive playlist: ruclips.net/p/PL8eNk_zTBST_SS_czCz6Do1yrUowhKBHI
i wonder if netflix carefully uses the metrics because my feedback is that their recommendations always sucks
Absolutely, it can be really frustrating when the recommendations don't seem to align with our tastes. Netflix uses complex algorithms based on a variety of metrics, including viewing history, user ratings, and even time of day you watch. Sometimes, though, it feels like these don't capture our preferences accurately. Have you tried tweaking your profile or rating more shows and movies? It might help refine what's suggested to you. Also, it's interesting to think about how different users experience these systems differently.
this entire second session was just a recap of the first one... literally NOTHING NEW in this one