Running process and method calling convention
There are a few parts of the logic that you need to be familiar with.
Fit command
In the fit
command, when partial_fit=f
(default), first the algorithm's fit
method is called. If it returns a DataFrame, we return that to the search. If it doesn't, we will call apply
later, then return whatever DataFrame apply
gives us.
Fit command partial_fit=t
In the fit
command, when partial_fit=t
, the algorithm's partial_fit
method is called on each chunk of 50,000 events. Then apply
is called on those events. We continue this process until we run out of data.
Apply Command
In the apply
command, things are much simpler. We reconstruct the python object by using its codec, then call apply
on the events, chunk by chunk.
Summary Command
In the summary
command, things are similar to apply
. We reconstruct the python object by using its codec, then call summary
.
Write an algorithm class | Correlation Matrix |
This documentation applies to the following versions of Splunk® Machine Learning Toolkit: 2.2.0, 2.2.1
Feedback submitted, thanks!