i pass "do_classification: True" parameter to ModelBuilders/gbm. no error. but it's not echoed and the result is regression

Description

is do_classification no longer used? there's no error if I use it
I'm getting a regression here when I expect logistic. Do I have to force output to enum? If so, there still should be error if I use a non-supported parameter (do_classification)

it's also not clear how to tell in the json response whether it got done as regression or logistic..what do I look at to tell?
(I'm not getting a cm in compute model metrics, but I think that's why)

POST , URI: /3/ModelBuilders.json/gbm, route: /3/ModelBuilders/gbm, parms: {validation_frame=covtype.train.hex, response_column=C55, max_depth=10, score_each_iteration=True, nbins=40, training_frame=covtype.train.hex, destination_key=covtype_gbm.hex, learn_rate=0.2, ntrees=2, min_rows=3, do_classification=True}
02-26 15:40:53.902 172.16.2.222:54321 9892 FJ-0-1 INFO: Building H2O GBM model with these parameters:
02-26 15:40:53.902 172.16.2.222:54321 9892 FJ-0-1 INFO: {"_destination_key":{"name":"covtype_gbm.hex","type":"Key"},"_train":{"name":"covtype.train.hex","type":"Key"},"_valid":{"name":"covtype.train.hex","type":"Key"},"_ignored_columns":null,"_dropNA20Cols":false,"_dropConsCols":true,"_score_each_iteration":false,"_max_confusion_matrix_size":20,"_response_column":"C55","_convert_to_enum":false,"_balance_classes":false,"_max_after_balance_size":5.0,"_class_sampling_factors":null,"_max_hit_ratio_k":10,"_ntrees":2,"_max_depth":10,"_min_rows":3,"_nbins":40,"_score_each_iteration":true,"_seed":0,"_checkpoint":false,"_loss":"AUTO","_learn_rate":0.2}
02-26 15:40:54.218 172.16.2.222:54321 9892 FJ-0-1 INFO: ==============================================================
02-26 15:40:54.218 172.16.2.222:54321 9892 FJ-0-1 INFO: r2 is 1.8941423592933049E-6, with 0x1 trees (average of 0.0 nodes)
02-26 15:40:56.756 172.16.2.222:54321 9892 FJ-0-1 INFO: 1. tree was built in 00:00:02.389 (Wall: 26-Feb 15:40:56.754)
02-26 15:40:56.999 172.16.2.222:54321 9892 FJ-0-1 INFO: ==============================================================
02-26 15:40:57.004 172.16.2.222:54321 9892 FJ-0-1 INFO: r2 is 0.22114617025935301, with 1x1 trees (average of 716.0 nodes)
02-26 15:40:58.915 172.16.2.222:54321 9892 FJ-0-1 INFO: 2. tree was built in 00:00:01.851 (Wall: 26-Feb 15:40:58.915)
02-26 15:40:59.187 172.16.2.222:54321 9892 FJ-0-1 INFO: ==============================================================
02-26 15:40:59.191 172.16.2.222:54321 9892 FJ-0-1 INFO: r2 is 0.3730283529899323, with 2x1 trees (average of 788.0 nodes)
02-26 15:40:59.521 172.16.2.222:54321 9892 # Session INFO: Method: GET , URI: /3/Models.json/covtype_gbm.hex, route: /3/Models/(?<key>.*), parms: {key=covtype_gbm.hex, find_compatible_frames=False}
02-26 15:40:59.568 172.16.2.222:54321 9892 # Session INFO: Method: GET , URI: /3/Models.json/covtype_gbm.hex, route: /3/Models/(?<key>.*), parms: {key=covtype_gbm.hex, find_compatible_frames=False}
02-26 15:40:59.580 172.16.2.222:54321 9892 # Session INFO: Method: GET , URI: /3/Frames.json/covtype.train.hex, route: /3/Frames/(?<key>.*), parms: {len=5, key=covtype.train.hex, offset=0, find_compatible_models=0}
02-26 15:40:59.601 172.16.2.222:54321 9892 # Session INFO: Method: POST , URI: /3/ModelMetrics.json/models/covtype_gbm.hex/frames/covtype.train.hex, route: /3/ModelMetrics/models/(?<model>.)/frames/(?<frame>.), parms: {model=covtype_gbm.hex, frame=covtype.train.hex}

Activity

Show:
Former user
March 14, 2015, 1:07 AM

do_classification and convert_to_enum are completely removed now. Thanks go to for all his work.

Fixed
Your pinned fields
Click on the next to a field label to start pinning.

Assignee

Former user

Reporter

Kevin Normoyle

CustomerVisible

No