Method for access to the current prediction for both the train and test set about xgboost HOT 2 CLOSED

predict take advantage of caching when you cache the DMatrix's prediction.
xgb.train and xgboost automatically cache results for train and all DMatrix
in watchlist.
So in customized obj, when you call predict, this is actually already what
you want it to be, it is optimized to fetch the buffered result, instead of
running prediction all over again

Tianqi

On Wed, Sep 24, 2014 at 11:55 AM, Blind Ape [email protected]
wrote:

From the wiki:
'The buffers are used to save the prediction results of last boosting step'

Would be very interesting a method for access to the current prediction
for both the train and test set.
This could speed up the xgb.iter.update function when using customized obj
(the xgb.predict call each iteration) and would let to estimate micro CV
error (merging cv predictions for each fold and compute the error metric
based in the 'honest' join prediction).

—
Reply to this email directly or view it on GitHub
https://github.com/tqchen/xgboost/issues/84.

from xgboost.

Great!

from xgboost.