85 Replies to “Frequent Pattern (FP) growth Algorithm for Association Rule Mining

  1. Thanks for the explanation, one of the more complete videos about this.
    Just a question, should the minimum support not be averaged with the number of transactions? I mean you can assume it means 3 transactions, but for me it usually meant 30%. In this case it would be 0.3*5 (nr of transactions) = 1.5 – rounds up to 2 transactions minimum.

  2. Thank you for this explanation! I have one advice, when you draw KEO on fp tree you are unnecessary drawing 'O' second time on the right side. You don't have to, because you have already linkage KEO on the left.

  3. support count for O is 4 not 3 … check once or twice while u r uploading a video …. dont put blindly nd make us fools

  4. StudyKorner, I like the video and the way the conditional FP-trees are got. However the common items in the conditional pattern base should be based on frequency count too ie if the frequency of a single item is more than 3(our min_sup here) then it should considered in the conditional FP-tree as well whether it is common or not. Cheers!!

  5. dude you should reduce the number of ads!! in a video of twenty minutes two ads are more than enough!otherwise a nice video.keep up the good work

  6. Good explanation, but can you please not use your finger to clean the board, it makes the board look really unclean, and you had 2 versions of your FP tree, it's kind of confusing, please keep it unambiguous.

  7. Please learn basic english first….or make the video in some other language you are comfortable with !!!!!

  8. you should not create new branch for O in the tree after K – E (for K-E-O). Thats why at the end it counts 4 instead of 3

  9. Building Frequent pattern from condition FP Tree after minute 22.19 in the case {KE:3}, it seems that one has to expand to power sets of {KE}. If that was {KEM} the non trivial power set would be {KO, EO, MO, KEO, KMO, EMO, KEMO,O}. Enumerating the powerset of a set is exponential on the number of items in the set. This in practical life that would not exceed 10. 2^10 = 1024 no big deal in computations.

  10. Clear explanation but lacks clarity in case of Conditional fp tree. One would not be able to understand clearly if tries to apply the same to the example in Introduction to Data Mining (Second Edition) by Micheal Steinback.

  11. for conditional FP Tree, you can't just take the common items from the Conditional Base pattern. You have to iterate the FP Growth algorithm again for each set of transactions in Conditional Base Pattern.

  12. Just one question, while making the FP free why did we find the common thing occurring in in the conditional database of the item? Is this always done?

  13. Error at 17:10
    Tree is already existing for k, e, o (o is at second lane ) plz.. Check
    No need to make new.
    (Thank you)

  14. At 20:28 for M in conditional pattern it should be {KE:2} not {KC:2} please do correct..The explanation is good.

  15. the conditional fp tree confuses me. There's literally like one example on the internet, and its like everywhere they do it differently. >.<

  16. What's the final step… We have to select rules based on the confidence as well? Or what u did in the video is final?

Leave a Reply

Your email address will not be published. Required fields are marked *