Hi,
Anyone here tried or have implemented AutoML features or meta learning features so Splunk can actually provide you a suggestion with few of ML algorithms and parameters based on your dataset?
I have been playing with MLTK, but still people is struggling with selecting algors and tuning hyper params.
Appreciate if anyone can shed a light in this ...
Hi,
Have you checked out the new field select option in the latest ML splunk release?
https://www.youtube.com/watch?v=SIHpKBMQmEQ
Now, this DOES NOT (and I do not think any tool, be it R/SAS/Python) can actually tell / make an educated guess on the algorithm you need to choose for making predictions. That is the pure stats part,however what the newly introduced field selector does do is make available for the user - which amongst the selected fields in a chosen algorithm is / are better suited and which all we can drop from consideration.
Maybe, if you give us an use case with some mock data we can help a bit more?
Hi,
Have you checked out the new field select option in the latest ML splunk release?
https://www.youtube.com/watch?v=SIHpKBMQmEQ
Now, this DOES NOT (and I do not think any tool, be it R/SAS/Python) can actually tell / make an educated guess on the algorithm you need to choose for making predictions. That is the pure stats part,however what the newly introduced field selector does do is make available for the user - which amongst the selected fields in a chosen algorithm is / are better suited and which all we can drop from consideration.
Maybe, if you give us an use case with some mock data we can help a bit more?