a8e24ef | jake-westfall | 01 March 2017, 20:05:15 UTC | Fixed typo | 01 March 2017, 20:05:15 UTC |
ca14031 | jake-westfall | 01 March 2017, 19:44:54 UTC | Update figures | 01 March 2017, 19:44:54 UTC |
98f2101 | jake-westfall | 08 January 2017, 17:57:39 UTC | Update paper link to point to Wellcome | 08 January 2017, 17:57:39 UTC |
208b503 | Tal Yarkoni | 23 November 2016, 06:25:22 UTC | Create LICENSE | 23 November 2016, 06:25:22 UTC |
33dc9b6 | Tal Yarkoni | 23 November 2016, 01:01:06 UTC | Update README.md | 23 November 2016, 01:01:06 UTC |
f7f4056 | jake-westfall | 27 September 2016, 03:08:25 UTC | Add link to preprint in README | 27 September 2016, 03:08:25 UTC |
f994e36 | jake-westfall | 22 September 2016, 22:02:08 UTC | Add Figures2to6 notebook; update README; correct typo in dir name | 22 September 2016, 22:02:08 UTC |
1bb2b84 | jake-westfall | 22 September 2016, 20:26:21 UTC | Added link to secondary OSF repository for 2 large files | 22 September 2016, 20:26:21 UTC |
469d0d4 | jake-westfall | 22 September 2016, 17:27:36 UTC | Updated links in README | 22 September 2016, 17:27:36 UTC |
326be46 | jake-westfall | 22 September 2016, 05:59:33 UTC | Minor tweaks to README | 22 September 2016, 05:59:33 UTC |
79f20e5 | jake-westfall | 22 September 2016, 05:46:51 UTC | Updated README | 22 September 2016, 05:46:51 UTC |
9cc9e7e | jake-westfall | 22 September 2016, 04:29:12 UTC | Add OpenfMRI emotion regulation data | 22 September 2016, 04:29:12 UTC |
b63d1e0 | jake-westfall | 21 September 2016, 21:42:50 UTC | Add NeuroVault/IAPS data and emotion stimulus mappings | 21 September 2016, 21:42:50 UTC |
3bb5e52 | jake-westfall | 21 September 2016, 19:52:00 UTC | Add analysis scripts, config files, emotion stims | 21 September 2016, 19:52:00 UTC |
db79906 | jake-westfall | 21 September 2016, 04:21:22 UTC | Merge branch 'master' of https://github.com/tyarkoni/nipymc | 21 September 2016, 04:21:22 UTC |
c8dbd73 | jake-westfall | 21 September 2016, 04:15:19 UTC | Add summary result files, simulation code, some figure code | 21 September 2016, 04:15:19 UTC |
6bab6bb | tyarkoni | 20 August 2016, 04:33:31 UTC | separate convolution for each run | 20 August 2016, 04:33:31 UTC |
bce0bd7 | tyarkoni | 19 August 2016, 21:35:53 UTC | Merge branch 'master' of github.com:tyarkoni/nipymc | 19 August 2016, 21:35:53 UTC |
0a04996 | tyarkoni | 19 August 2016, 21:35:46 UTC | fixed convolution bug | 19 August 2016, 21:35:46 UTC |
4956757 | Tal Yarkoni | 14 August 2016, 21:39:43 UTC | Merge pull request #16 from jake-westfall/master Fix level_map issue | 14 August 2016, 21:39:43 UTC |
2968e1e | jake-westfall | 14 August 2016, 20:22:48 UTC | Fix level_map issue When adding a fixed term that has already been indirectly referenced in a previous call to add_term (for example, through the split_by or orthogonalize arguments), the level mapping for the fixed term was never getting added to level_map. (But everything worked fine if the fixed term was added BEFORE ever getting referenced in a split_by or orthogonalize argument.) It turns out that this can be fixed by a fairly simple tweak to _get_variable_data, namely, assigning default labels to all 'variable' arguments and hashing labels rather than variables. | 14 August 2016, 20:22:48 UTC |
d459aba | Tal Yarkoni | 20 May 2016, 05:41:40 UTC | Merge pull request #15 from jake-westfall/master Fix bug causing user-supplied start values to be ignored | 20 May 2016, 05:41:40 UTC |
adec316 | jake-westfall | 20 May 2016, 05:24:42 UTC | Fix bug causing user-supplied start values to be ignored model.run() included a named 'start' parameter for passing to the pymc3 sampler, but also had code that assumed 'start' would be part of kwargs, with the result that user-supplied start values were always overwritten by default values. This commit simply removes 'start' from the method signature, so it will always be handled as a kwarg instead. | 20 May 2016, 05:24:42 UTC |
21cff6d | Tal Yarkoni | 09 May 2016, 00:16:18 UTC | Merge pull request #14 from tyarkoni/dm-plotting Dm plotting | 09 May 2016, 00:16:18 UTC |
2b2141d | tyarkoni | 09 May 2016, 00:13:57 UTC | fixes row labels for different DM panels | 09 May 2016, 00:13:57 UTC |
9f1d657 | tyarkoni | 09 May 2016, 00:07:15 UTC | better DM plotting at construction time | 09 May 2016, 00:07:15 UTC |
039f9c8 | Tal Yarkoni | 05 May 2016, 18:32:40 UTC | Merge pull request #11 from tyarkoni/language-task Added HCP language task support | 05 May 2016, 18:32:40 UTC |
5e96904 | tyarkoni | 05 May 2016, 18:32:12 UTC | Added HCP language task support | 05 May 2016, 18:32:12 UTC |
92ef1fd | Tal Yarkoni | 05 May 2016, 18:30:42 UTC | Merge pull request #10 from tyarkoni/interpolation adds proper interpolation; closes #6 | 05 May 2016, 18:30:42 UTC |
ae4c354 | tyarkoni | 05 May 2016, 18:29:44 UTC | resolved conflict | 05 May 2016, 18:29:44 UTC |
abce21b | tyarkoni | 05 May 2016, 18:21:42 UTC | adds proper interpolation; closes #6 | 05 May 2016, 18:21:42 UTC |
2354fdf | Tal Yarkoni | 29 April 2016, 00:57:57 UTC | Merge pull request #8 from jake-westfall/master Allow passing 'chain' argument to pymc3.run | 29 April 2016, 00:57:57 UTC |
e451789 | jake-westfall | 28 April 2016, 23:50:00 UTC | Allow passing 'chain' argument to pymc3.run Normally one will run multiple parallel MCMC chains by just setting njobs > 1. But in some cases the parallelization must be done manually (e.g., TACC doesn't seem to like it when you use njobs > 1). In those cases you can run individual chains and then merge them later into a MultiTrace object, provided that the chains have unique chain IDs. The 'chain' argument is used to set the chain IDs of individual chains. | 28 April 2016, 23:50:00 UTC |
15295c9 | Tal Yarkoni | 28 April 2016, 21:30:48 UTC | Merge pull request #7 from tyarkoni/subject-level-ar model separate AR terms for each scanning run | 28 April 2016, 21:30:48 UTC |
3204371 | tyarkoni | 28 April 2016, 21:28:50 UTC | model separate AR terms for each scanning run | 28 April 2016, 21:28:50 UTC |
6a7164e | tyarkoni | 26 April 2016, 14:44:07 UTC | fixes design matrix initialization for continuous variables | 26 April 2016, 14:44:07 UTC |
de655c2 | tyarkoni | 25 April 2016, 23:54:44 UTC | Merge branch 'master' of github.com:tyarkoni/nipymc | 25 April 2016, 23:54:44 UTC |
29e8d1a | tyarkoni | 25 April 2016, 23:54:16 UTC | enable convolution skipping | 25 April 2016, 23:54:16 UTC |
afbaf02 | tyarkoni | 25 April 2016, 23:52:45 UTC | fixes bug in results summarization | 25 April 2016, 23:52:45 UTC |
17c270b | tyarkoni | 25 April 2016, 23:52:22 UTC | _get_variable_data returns copy of DM; prevents double convolution | 25 April 2016, 23:52:22 UTC |
de832d5 | Tal Yarkoni | 22 April 2016, 21:27:23 UTC | Merge pull request #5 from jake-westfall/master Add `estimate_random_mean` to add_term() | 22 April 2016, 21:27:23 UTC |
2d5d22e | jake-westfall | 19 April 2016, 20:06:33 UTC | estimate_random_mean now works when split_by=None | 19 April 2016, 20:06:33 UTC |
868923e | jake-westfall | 19 April 2016, 19:21:53 UTC | Add 'estimate_random_mean' to add_term() If False (default), set mean of random effect distribution to 0. If True, estimate a mean parameter (in which case the corresponding fixed effect should be omitted, for identifiability reasons). Allows user to fit models parameterized as in the hierarchical model at http://twiecki.github.io/blog/2014/03/17/bayesian-glms-3/ | 19 April 2016, 19:21:53 UTC |
ee1f4fc | tyarkoni | 31 March 2016, 19:16:24 UTC | save level-specific mapping | 31 March 2016, 19:16:24 UTC |
231e4ab | tyarkoni | 31 March 2016, 18:48:52 UTC | save level --> index mapping in model | 31 March 2016, 18:48:52 UTC |
9addbf7 | tyarkoni | 31 March 2016, 18:40:14 UTC | simplified data extraction logic | 31 March 2016, 18:40:14 UTC |
9f53850 | tyarkoni | 31 March 2016, 17:45:17 UTC | Merge branch 'master' of github.com:tyarkoni/nipymc Conflicts: nipymc/model.py | 31 March 2016, 17:45:17 UTC |
188d520 | tyarkoni | 31 March 2016, 17:42:59 UTC | plotting handles both terms and contrasts | 31 March 2016, 17:42:59 UTC |
1df0041 | Tal Yarkoni | 23 March 2016, 16:49:57 UTC | Merge pull request #4 from jake-westfall/master add_term accepts list of variables | 23 March 2016, 16:49:57 UTC |
3808997 | jake-westfall | 23 March 2016, 06:55:29 UTC | move 'burn' parameter to summarize() | 23 March 2016, 06:55:29 UTC |
8b3f72f | jake-westfall | 23 March 2016, 06:33:58 UTC | add_term accepts list of variables | 23 March 2016, 06:33:58 UTC |
11dc4ca | Tal Yarkoni | 22 March 2016, 22:11:25 UTC | Merge pull request #3 from jake-westfall/master Add unique stimulus face IDs to EmotionTaskFactory | 22 March 2016, 22:11:25 UTC |
8f9216e | jake-westfall | 22 March 2016, 21:56:32 UTC | Add unique stimulus face IDs to EmotionTaskFactory The 'left-stim', 'top-stim', and 'right-stim' columns now contain the unique stimulus face identities, rather than the previous labels such as MF2_left. The 'stimulus' column contains the old labels, and thus only identifies whole configurations of stimuli. Note that 'repetitions' only tracks how many times the 'stimulus' has been repeated, not how many times 'left-stim' etc. have been repeated. | 22 March 2016, 21:56:32 UTC |
b6c3837 | tyarkoni | 15 March 2016, 16:25:32 UTC | adds ability to easily add Deterministics via eval | 15 March 2016, 16:25:32 UTC |
35a348c | tyarkoni | 14 March 2016, 20:03:02 UTC | fixes broken convolution kwargs | 14 March 2016, 20:03:02 UTC |
39dea55 | tyarkoni | 14 March 2016, 19:57:44 UTC | adds rudimentary Dataset testing | 14 March 2016, 19:57:44 UTC |
2310aef | tyarkoni | 14 March 2016, 19:57:04 UTC | cleanup | 14 March 2016, 19:57:19 UTC |
da9af69 | tyarkoni | 14 March 2016, 19:56:28 UTC | updated setup.py and requirements | 14 March 2016, 19:56:28 UTC |
5c13a53 | tyarkoni | 14 March 2016, 19:14:18 UTC | adds basic tests for convolutions module | 14 March 2016, 19:14:18 UTC |
1b89011 | tyarkoni | 14 March 2016, 19:02:23 UTC | default to no kwargs in get_convolution | 14 March 2016, 19:09:31 UTC |
c26d2f9 | tyarkoni | 14 March 2016, 18:49:49 UTC | model run returns BayesianModelResults instance | 14 March 2016, 18:49:49 UTC |
a48c364 | Tal Yarkoni | 14 March 2016, 05:28:04 UTC | Merge pull request #1 from jake-westfall/master Correctly set n_vols when directly creating Dataset | 14 March 2016, 05:28:04 UTC |
2eb7ceb | jake-westfall | 14 March 2016, 03:51:15 UTC | Correctly set n_vols when directly creating Dataset When creating a Dataset object directly (rather than using HCPDatasetFactory), n_vols was initialized as the max() of the 'vol' column in the activation dataset, but this only correctly counts n_vols if the vols are indexed from 1. This commit initializes n_vols as the nunique() of 'vol', which works in the 0- or 1-indexed cases. | 14 March 2016, 03:51:15 UTC |
63ff50c | tyarkoni | 09 March 2016, 20:29:17 UTC | moved README | 09 March 2016, 20:29:17 UTC |
680d3f5 | tyarkoni | 09 March 2016, 20:28:27 UTC | name change | 09 March 2016, 20:28:27 UTC |
6b50988 | tyarkoni | 09 March 2016, 20:27:30 UTC | initial commit | 09 March 2016, 20:27:30 UTC |