Age | Commit message (Collapse) | Author |
|
|
|
|
|
CSV: order of rows in order of compounds
fminer check_params: auto-detecting prediction feature in case of a
single feature in the dataset
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Conflicts:
lib/compound.rb
|
|
|
|
|
|
Only merged lib/compound.rb
|
|
Upstream fix from master branch: Bugfix. Lookup problem for external
compound uri fixed.
Now external hostname/port in compound uri doesn't influence lookup in
external feature datasets.
|
|
|
|
Now external hostname/port in compound uri doesn't influence lookup in external feature datasets.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
add k-fold-split method for crossvalidation, add sammon plotting
|
|
|
|
|
|
row names
|
|
|
|
Moved back to setting with best performance (see table 18,
http://goo.gl/uGMJN).
|
|
|
|
train_succeess and dataset cleanup + own chances (get compound name
function, small dataset fix)
|
|
|
|
train() failure for censoring predictions (algorithm.rb),
transformation log for creating n_prop and q_prop fixed (transform.rb)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
(del_nomnial is false by default).
|
|
|