Fine tutorial, however, there's underlying issue with k-means, that it will always split data into k clusters even though you'd actually have less than k clusters in your data. Please consider DBSCAN as an alternative :)
It might be ok to use k-means clustering for anomaly detection but, as mentioned in your video, there are things we need to think over about the threshold to tell it is an anomaly and the optimal number of clusters. Finding the number of clusters using Elbow method is not really applicable to the real time situation because we don't really know the optimal number of clusters unless we really do the data mining to find it.
Hi you are correct - unless you have very well explored domain problem and know how many clusters to expect, k-means may be difficult to apply in more complex areas. It's mostly used for data exploration.
Thanks for this video and tutorial, please can someone try this on Matlab or Devc++ Applications? Then this is related to my work can you please share the codes with me privately? I will so much appreciate it. Thanks once again.
Fine tutorial, however, there's underlying issue with k-means, that it will always split data into k clusters even though you'd actually have less than k clusters in your data. Please consider DBSCAN as an alternative :)
It might be ok to use k-means clustering for anomaly detection but, as mentioned in your video, there are things we need to think over about the threshold to tell it is an anomaly and the optimal number of clusters. Finding the number of clusters using Elbow method is not really applicable to the real time situation because we don't really know the optimal number of clusters unless we really do the data mining to find it.
Hi you are correct - unless you have very well explored domain problem and know how many clusters to expect, k-means may be difficult to apply in more complex areas. It's mostly used for data exploration.
How to find threshold for detecting outliers?
Hello bro can you please make a video on real time forecasting on live stock market
why have you not used SVMModel in Apache spark to create the model? is it because of no of data is too large?
Great!
Be great to access the code.
nice technique, is it possible to share the code Please...
Thanks for this video and tutorial, please can someone try this on Matlab or Devc++ Applications? Then this is related to my work can you please share the codes with me privately? I will so much appreciate it. Thanks once again.
Great tutorial. please share code.
Great tutorial. please share code :)
can you send the code to me, please?
my nigga mariusz
can you send the code to me, please?