|
|
|
@288
|
[288]
|
05/14/11 12:21:49 |
dsimcha |
Minor updates for 2.053. |
|
|
|
@286
|
[286]
|
03/12/11 11:27:44 |
dsimcha |
Fix singularness stuff. |
|
|
|
@285
|
[285]
|
03/07/11 22:22:06 |
dsimcha |
This should be even better numerically. |
|
|
|
@284
|
[284]
|
03/07/11 20:07:05 |
dsimcha |
This should help w/ numerical stability a little. |
|
|
|
@283
|
[283]
|
03/04/11 22:38:40 |
dsimcha |
Should be the last of the ridge regression numerical issues. |
|
|
|
@282
|
[282]
|
03/04/11 09:30:28 |
dsimcha |
This should fix some of the convergence failures related to oscillation … |
|
|
|
@281
|
[281]
|
03/03/11 20:38:16 |
dsimcha |
These should use relative error. This also fixes some convergence issues. |
|
|
|
@279
|
[279]
|
03/03/11 18:37:29 |
dsimcha |
Fix convergence issues. |
|
|
|
@273
|
[273]
|
02/19/11 18:45:07 |
dsimcha |
Work around compiler bugs. |
|
|
|
@271
|
[271]
|
02/19/11 10:29:54 |
dsimcha |
...and don't leak memory while doing it. |
|
|
|
@270
|
[270]
|
02/19/11 10:28:48 |
dsimcha |
Fall back to solving directly in pathological cases. |
|
|
|
@269
|
[269]
|
02/15/11 21:06:02 |
dsimcha |
Recycle starting values. |
|
|
|
@267
|
[267]
|
02/13/11 00:41:10 |
dsimcha |
Lasso and efficient p >> n ridge regression. |
|
|
|
@266
|
[266]
|
02/12/11 19:18:03 |
dsimcha |
Mostly WIP, some trivial fixes. |
|
|
|
@265
|
[265]
|
02/10/11 21:58:30 |
dsimcha |
WIP |
|
|
|
@264
|
[264]
|
02/08/11 23:26:22 |
dsimcha |
Use Cholesky decomposition instead of Gaussian elimination. About 3x … |
|
|
|
@263
|
[263]
|
02/08/11 08:45:30 |
dsimcha |
Minor fixes and work in progress. |
|
|
|
@258
|
[258]
|
02/04/11 23:02:13 |
dsimcha |
Speed up matrix ops for regression. |
|
|
|
@257
|
[257]
|
02/04/11 20:32:43 |
dsimcha |
Some obvious in hindsight cache optimizations. |
|
|
|
@256
|
[256]
|
02/04/11 08:59:12 |
dsimcha |
Duh...Better yet, go with the grain for better cache performance. |
|
|
|
@255
|
[255]
|
02/04/11 08:46:14 |
dsimcha |
Optimization: Compute half the matrix, then symmetrize, rather than … |
|
|
|
@254
|
[254]
|
02/03/11 00:24:05 |
dsimcha |
One more bug fix. |
|
|
|
@253
|
[253]
|
02/02/11 23:42:36 |
dsimcha |
Fix a major and amazingly well hidden memory recycling bug. |
|
|
|
@252
|
[252]
|
02/02/11 22:55:22 |
dsimcha |
Add logistic ridge, redesign linear ridge. |
|
|
|
@251
|
[251]
|
02/01/11 23:32:38 |
dsimcha |
Don't intermingle ridge w/ logistic. |
|
|
|
@249
|
[249]
|
02/01/11 23:27:23 |
dsimcha |
Add ridge regression, a few small misc, cleanups. |
|
|
|
@248
|
[248]
|
01/27/11 09:40:43 |
dsimcha |
Fix some minor silliness. |
|
|
|
@247
|
[247]
|
01/26/11 21:41:04 |
dsimcha |
docs |
|
|
|
@246
|
[246]
|
01/26/11 21:33:53 |
dsimcha |
Add full hypothesis testing for logistic regression. |
|
|
|
@245
|
[245]
|
01/26/11 09:39:19 |
dsimcha |
Fix NaN/degenerate case handling in logisticRegressBeta. |
|
|
|
@236
|
[236]
|
01/08/11 09:37:37 |
dsimcha |
Improve error messages. |
|
|
|
@233
|
[233]
|
12/13/10 19:51:00 |
dsimcha |
Exception vs. NaN convention. |
|
|
|
@226
|
[226]
|
11/13/10 18:58:27 |
dsimcha |
Fix 64-bit compilation errors. Code only passes semantic analysis, hasn't … |
|
|
|
@216
|
[216]
|
09/09/10 20:37:16 |
dsimcha |
Residuals should copy input ROR to avoid unexpected side effects. |
|
|
|
@211
|
[211]
|
08/22/10 21:53:08 |
dsimcha |
Fix ranges: Add save() and @property stuff. |
|
|
|
@209
|
[209]
|
08/22/10 21:23:02 |
dsimcha |
Fix some style issues and a few 64 issues. |
|
|
|
@190
|
[190]
|
07/29/10 01:18:15 |
dsimcha |
Argh, a few more small fixes. |
|
|
|
@189
|
[189]
|
07/29/10 00:53:38 |
dsimcha |
Fix terminology. |
|
|
|
@188
|
[188]
|
07/29/10 00:24:15 |
dsimcha |
Improve docs a little. |
|
|
|
@186
|
[186]
|
07/29/10 00:17:28 |
dsimcha |
Add logistic regression, though it remains to be fleshed out. |
|
|
|
@182
|
[182]
|
07/18/10 21:57:02 |
dsimcha |
Add Unqual in a few places. |
|
|
|
@181
|
[181]
|
07/18/10 21:49:39 |
dsimcha |
Added linearRegressBuf and polyFitBuf to allow explicit buffers. |
|
|
|
@180
|
[180]
|
07/15/10 00:48:07 |
dsimcha |
Revert changes for now b/c I'm REALLY not sure they're right. |
|
|
|
@179
|
[179]
|
07/14/10 22:42:37 |
dsimcha |
Removed predict() because I realized it doesn't cover enough cases to be … |
|
|
|
@178
|
[178]
|
07/14/10 22:17:01 |
dsimcha |
Bugfix/improve dstats.regress. |
|
|
|
@174
|
[174]
|
07/11/10 23:07:50 |
dsimcha |
Add save for Residuals. |
|
|
|
@164
|
[164]
|
06/19/10 18:23:09 |
dsimcha |
Add AUROC, misc. 2.047-related updates. |
|
|
|
@145
|
[145]
|
02/28/10 22:49:52 |
dsimcha |
80-bit reals -> 64-bit doubles, since this results in up to 3x performance … |
|
|
|
@132
|
[132]
|
01/09/10 10:02:11 |
dsimcha |
Added paritial correlation, residuals. Finish renaming chiSqr to … |
|
|
|
@131
|
[131]
|
12/27/09 20:58:10 |
dsimcha |
More error handling improvements. |
|
|
|
@119
|
[119]
|
11/30/09 22:47:08 |
dsimcha |
Rename correlation functions to more descriptive names. Old names are … |
|
|
|
@113
|
[113]
|
11/22/09 12:45:01 |
dsimcha |
Add adjustedR2 to regress, misc. small fixes. |
|
|
|
@88
|
[88]
|
10/18/09 06:57:55 |
dsimcha |
Misc. code cleanup, refactoring. |
|
|
|
@75
|
[75]
|
05/22/09 19:57:44 |
dsimcha |
Workaround for Bug 3006. |
|
|
|
@60
|
[60]
|
05/14/09 23:10:56 |
dsimcha |
Misc. cleanup and optimization. |
|
|
|
@52
|
[52]
|
05/09/09 13:11:19 |
dsimcha |
New module: linear regression. Also, a few doc fixes. |