Splunk Search

MLTK LogisticRegression probabilities=true only showing first event probability

eduardoduarte
Explorer

I Have trained a LogisiticRegression model by using TFIDF data (3K events in a month) as input successfully using probabilities=true

In the fit process it shows the probabilities of everything correctly, I can even do a ROC curve analysis. 

The problem comes when use the model by doing a new search and TFIDF the data, and right after the  "|apply logistic_model probabilities=true"  to new data (say... last 24 hours). The behavior is that it only shows the probabilities for the first event (sometimes two or three but not all if I apply the model to "old data") and the others appear blank but the predicted field appears correctly.

Now, if I do a search and I apply only the TFIDF_model, without the apply logistic_model and then I "|loadjob  123ABC"  having only the TFIDF data calculated previously and then  Iapply the model to the loaded job of TFIDF data, the probabilities appear magically.

I am almost sure this is a bug, but I want to know if there is some workaround ?

 

Thanks

0 Karma
Career Survey
First 500 qualified respondents will receive a $20 gift card! Tell us about your professional Splunk journey.

Can’t make it to .conf25? Join us online!

Get Updates on the Splunk Community!

Can’t Make It to Boston? Stream .conf25 and Learn with Haya Husain

Boston may be buzzing this September with Splunk University and .conf25, but you don’t have to pack a bag to ...

Splunk Lantern’s Guide to The Most Popular .conf25 Sessions

Splunk Lantern is a Splunk customer success center that provides advice from Splunk experts on valuable data ...

Unlock What’s Next: The Splunk Cloud Platform at .conf25

In just a few days, Boston will be buzzing as the Splunk team and thousands of community members come together ...