gitlog.txt

# git log for private repository github.com/umncs-caringbridge/recsys-peer-match
# Generated on Sep 14 2022 using `git log --pretty=format:"%as %h %an %s"`
2022-06-29 84cada9 Zachary Levonian Added coverage analysis for second design requirement
2022-06-27 5a6ec3e Zachary Levonian Added matrix factorization baseline
2022-06-23 f57697d zentavious footnote analysis - churn
2022-06-22 aabf4ef zentavious Including second visits in Sample size calculations
2022-06-21 e748c9c zentavious redid effect sizes with ne parameters
2022-06-20 9619e2a Zachary Levonian Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2022-06-20 72e8e9e Zachary Levonian Compute percentage
2022-06-17 555b0e7 zentavious revised effect size calcs
2022-06-15 dca210b zentavious b0 analysis too
2022-06-15 c09b08e zentavious full study rct calculations
2022-06-15 423406f zentavious recurring intervention effect size calcs for participants
2022-06-14 233b805 Zachary Levonian Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2022-06-14 aea3325 Zachary Levonian Added click data analysis, including some useful tables
2022-06-13 a2d8972 zentavious fig adjustment fig 11
2022-06-13 a503ffe zentavious resizing fig 13
2022-06-13 1fb46ba zentavious fix bug re-reccing in unfiltured group
2022-06-12 ed95281 Zachary Levonian Updated analysis notebook with feature ablations results.
2022-06-12 c8e337d Zachary Levonian Added analysis of validation results for feature ablations, confirming that a textless model does better.
2022-06-11 2d9f3bd Zachary Levonian Added baseline implementation, added to paper. Copied ZachEval notebook to create a copy of the LinearNet results
2022-06-10 9ec4673 Zachary Levonian Attempting to use the initiations already generated to generate the initated with baselines
2022-06-10 d0920e1 Zachary Levonian Adjustments to training configurations generated for feature ablations
2022-06-10 5171bff Zachary Levonian Add stub for new experiment using feature ablations.
2022-06-10 ce797ef Zachary Levonian Added MLP_study and MLP_tuned results.
2022-06-10 be1ef61 Zachary Levonian Added new annotation data and figure
2022-06-09 b7af18a Zachary Levonian Began analyzing the LinearNet training results, including 11 models that failed to finish training within 12 hours. Updates to script parameters/resource requests.
2022-06-09 cff36a9 Zachary Levonian Added additional modeling details, but also identified a need for 40 additional journal annotations
2022-06-09 0c1ebd6 Zachary Levonian Added stub notebook with simnet results for future reference
2022-06-09 a55a444 Zachary Levonian Checkpoint of notebook with Cosine Similarity / SimNet results.
2022-06-08 df41c53 Zachary Levonian Added new stub code for model eval, including a completed evaluation of the Cosine Similarity model. Added loading method to Model Manager that tracks the epoch with the best achieved validation loss.
2022-06-08 a0f3e1d zentavious adding totals
2022-06-08 4e55288 Zachary Levonian Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2022-06-08 8fa9a2b Zachary Levonian Added new model evaluation approach to deal with the inconsistent pickle loading times.
2022-06-08 1cec25d zentavious pre rec dataframe feather
2022-06-07 0bf9884 zentavious Redoing some figures
2022-06-03 b540522 Zachary Levonian Added notebooks computing some of the baselines. Still need to implement the ones that aren't contained in the non-text features
2022-06-03 81f0e11 Zachary Levonian no longer computing NDCG
2022-06-03 ef850d1 zentavious clipping
2022-06-03 076be7a zentavious excess weekly actions
2022-06-03 b637299 zentavious more fig adjusting
2022-06-03 c2705a0 zentavious adjusting appendix fig size
2022-06-03 e6d7fdc zentavious adjusting fig labels
2022-06-03 19dcc06 zentavious adjusting fig size
2022-06-03 bc9f355 zentavious fixed overflow allignment
2022-06-03 655d90b zentavious Merge branch 'main' of https://github.com/umncs-caringbridge/recsys-peer-match into main
2022-06-03 2903be8 zentavious Recced outcome estimates
2022-06-03 20cb092 Zachary Levonian Started running test and predict text feature generation
2022-06-03 88c2a0a zentavious Merge branch 'main' of https://github.com/umncs-caringbridge/recsys-peer-match into main
2022-06-03 81f9099 Zachary Levonian Small documentation updates. Added mistakenly-omitted figure.
2022-06-03 974462c Zachary Levonian Added script to recover when test_context generation gets corrupted. Updated gen_continue script and make_text_features_train scripts with agate defaults.
2022-06-02 7cfbfcb zentavious rename
2022-06-02 24082b7 zentavious recced outcome estimates checkpoint
2022-06-02 d33cf9d Zachary Levonian Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2022-06-02 1fbf82a Zachary Levonian Added analysis to generate survey figure.
2022-06-02 144fe05 zentavious amp interaction analysis
2022-06-02 fbcbcd2 Zachary Levonian Documentation updates, and extend eval scripts to specifically support splitting by validation/test period.
2022-06-01 5976251 Zachary Levonian Added an old email mockup; no idea what this is or if it was used.
2022-06-01 78494b5 Zachary Levonian Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2022-06-01 796b0ca Zachary Levonian Fixed a few bugs and added additional logging
2022-06-01 4381daa zentavious Merge branch 'main' of https://github.com/umncs-caringbridge/recsys-peer-match into main
2022-06-01 af4700c zentavious revert overriden figure
2022-05-31 d9b05d3 Zachary Levonian Added sidebar analysis on coverage
2022-05-31 746e335 Zachary Levonian Removed baseline scoring from recDataGenerator and tweaked script parameters, after a foolish attempt to recover from mismatches between previous and current journal snapshots.
2022-05-29 456fc63 Zachary Levonian Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2022-05-29 db260e0 Zachary Levonian Update survival analysis approach with new data, added summary table to paper
2022-05-28 4d910e8 zentavious post click analysis
2022-05-27 0d15f11 zentavious setting up for test sbatch
2022-05-27 3963002 zentavious Merge branch 'main' of https://github.com/umncs-caringbridge/recsys-peer-match into main
2022-05-27 ed43a3c zentavious pre-click analysis + control group bug fix
2022-05-27 1313ffc Zachary Levonian Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2022-05-27 88c3bfb Zachary Levonian Started updating survival analysis to use more recent data
2022-05-26 1656275 zentavious exploring duplicate control keys
2022-05-26 bd2f92c zentavious Pulling new df for all sites
2022-05-26 b9530ca zentavious On click comparison over time setup to run overnight
2022-05-25 a1a71dc zentavious Merge branch 'main' of https://github.com/umncs-caringbridge/recsys-peer-match into main
2022-05-25 dd4fce0 Zachary Levonian Computed percentiles for P1
2022-05-25 edcc66c zentavious apply vs join
2022-05-25 03c461a Zachary Levonian Add small demo of randomly sampling fake click times
2022-05-25 61b87e3 zentavious fixed rank/score by batch figs
2022-05-24 f3ce780 Zachary Levonian More figures and computing of counts.
2022-05-24 26921a1 zentavious discriptive stats tables
2022-05-24 c576052 zentavious 10 bootstraps rec comparison, click/rec/click_rec
2022-05-24 c483a78 zentavious 12 week analysis of psuedo recced comparison
2022-05-24 1c2ef39 Zachary Levonian Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2022-05-24 a3bd1e0 Zachary Levonian Added participant activity counts and exported click data for usage.
2022-05-23 a3ca407 zentavious setting up recced long comparisson
2022-05-23 465a750 zentavious ci_estimates for recced sites
2022-05-23 81c8600 zentavious Merge branch 'main' of https://github.com/umncs-caringbridge/recsys-peer-match into main
2022-05-23 df83ea5 zentavious Don't remember what this was, just commiting my personal files
2022-05-22 9263a40 Zachary Levonian Minor addition of comparison between a flexible pre-study window and the maximum pre-study window
2022-05-21 f54accc Zachary Levonian Added sensitivity analysis for time window selection.
2022-05-20 0caff89 Zachary Levonian Added code for robust SEs for the Poisson model and adjusted the generation and reporting of the U statistic
2022-05-20 e221bf1 zentavious rsquared analysis
2022-05-19 9cfd446 Zachary Levonian Chose mine, potentially losing changes
2022-05-19 09a5a36 Zachary Levonian Added a bunch of analysis, including bootstrapping estimates and additional variables
2022-05-19 57cf2aa zentavious forgot to save
2022-05-19 4770dd7 zentavious added recced site features, and began ols analysis
2022-05-19 c7ceb6b zentavious recced sites analysis checkpoint
2022-05-18 1df7060 Zachary Levonian Added revised flow to feature generation to fix a few bugs and explore a few data mysteries
2022-05-18 7506f59 Zachary Levonian Added code to produce a summary table for pre-study participant activity.
2022-05-17 ea9e2ca zentavious finished control features
2022-05-17 fe466cf Zachary Levonian Added working implementations of IP weighting, standardization, and doubly robust estimators.
2022-05-17 91e3e48 zentavious added n_self_site_interactedwith, time_since_first_journal_update, and is_prestudy flag
2022-05-16 49b6e2a Zachary Levonian Added an apparently broken implementation of IP weighting
2022-05-13 42620b1 Zachary Levonian Added additional analysis to notebook, from meeting with Matthew
2022-05-13 ff6dfe7 Zachary Levonian Added initial form of retention notebook
2022-05-13 dccfed6 Zachary Levonian Cleaned up notebook to extract pseudocontrol group.
2022-05-13 32c8f31 Zachary Levonian About to clean up EndOfRecruitmentSiteProfileSampling
2022-05-12 c5089a9 Zachary Levonian Redoing site_profile extraction with more permissive net, hoping to better explain the missing visit data among participants.
2022-05-12 691c1fd Zachary Levonian Added tentative Transformation interface and test log transform.
2022-05-05 58767c9 zentavious Analysis of batch score means
2022-05-05 f1dfa87 zentavious saved notebook
2022-05-05 b8e7285 zentavious Added csv of sites for analysis in root_dir/analysis/
2022-05-05 b395e5d zentavious Bug fix
2022-05-05 26d21a2 zentavious Saving figures
2022-05-03 2532aae zentavious analysis of factual, counterfactual, and control conditions
2022-05-03 1e4f6b5 zentavious Merge branch 'main' of https://github.com/umncs-caringbridge/recsys-peer-match into main
2022-05-03 e726e94 zentavious mean scores analysis
2022-05-02 29b449b Zachary Levonian Add stub CF implementation
2022-04-28 3eb5cd7 Zachary Levonian Update to docs
2022-04-28 8f50149 Zachary Levonian Documentation updates
2022-04-28 4db4764 Zachary Levonian Documentation updates, add option to specify minibatch size, add stub InteractionNet model.
2022-04-28 49da5e0 Zachary Levonian Updated documentation
2022-04-28 be2a224 Zachary Levonian Added stub model implementation building a learned form of simnet.
2022-04-27 2ec8106 Zachary Levonian Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2022-04-27 491361a Zachary Levonian Added new cloudfront data
2022-04-25 bc9448e zentavious Merge branch 'main' of https://github.com/umncs-caringbridge/recsys-peer-match into main
2022-04-25 b8bf7fa zentavious gathering control groups
2022-04-21 7515f1b Zachary Levonian Update documentation
2022-04-21 f3a5034 Zachary Levonian Updated with new data dump
2022-04-21 54c0259 Zachary Levonian Updated activity monitoring notebook with small example
2022-04-11 d7af4d7 Zachary Levonian Updated with newer data to refresh counts.
2022-04-01 523131b matthewmac99 Update evaluateModel.sh
2022-03-31 510ffdd matthewmac99 Increased memory to 20g
2022-03-31 b35482e Ngan Nguyen newest evaluation for concat
2022-03-30 d95dada Ngan Nguyen Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2022-03-30 3b49862 Ngan Nguyen added learning rate for concat
2022-03-29 99aa54a Zachary Levonian Added test config and fixed preprocessing
2022-03-29 8d312fa Zachary Levonian Fixed preprocessing?
2022-03-25 78b7c1e Ngan Nguyen Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2022-03-25 4b7a647 Ngan Nguyen added more learning rates to concat
2022-03-25 c94660d Zachary Levonian Fixed flag
2022-03-25 769c7c4 Zachary Levonian Added validation-only flag
2022-03-25 8b37f2d Zachary Levonian Added flag to make evaluation faster.
2022-03-25 77735a3 matthewmac99 Update evaluateModel.sh
2022-03-18 aafb756 Zachary Levonian Added code to plot validation loss during training
2022-03-18 07eac83 Matthew Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2022-03-18 01defdc Matthew adding adjusted feature ablation experiment
2022-03-18 ddeba66 Zachary Levonian Adjusted down the memory requirements for evaluation
2022-03-18 bfa1846 Ngan Nguyen concat net evaluations on full rec metadata
2022-03-11 9a74718 Zachary Levonian Added initial analysis of annotation data and click data, including code to create the IRR table and an initial estimate of the proportions of each annotated category.
2022-03-10 df04b99 Matthew Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2022-03-10 f43a3d1 Matthew adding AdamExperimentAnalysis
2022-03-10 b115489 Zachary Levonian Small bugfixes, including fixing stdout and stderr paths during eval
2022-03-10 c2ed257 Zachary Levonian Added new evaluation flow for experiments
2022-03-10 cc03c35 Ngan Nguyen Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2022-03-10 3d0bc23 Ngan Nguyen added changes in concatnet
2022-03-08 f543d91 Zachary Levonian Cosmetic and doc updates to script
2022-03-07 7849864 Ngan Nguyen added script to generate rec context pickle files
2022-02-23 16916b4 Ngan Nguyen Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2022-02-23 552f6fc Ngan Nguyen added concatnet training experiment
2022-02-23 b9d08eb Zachary Levonian Rewrite dyad extraction to focus on recommended dyads
2022-02-23 e40e899 Zachary Levonian Checkpoint before reorienting around user/site dyads
2022-02-17 f00bad3 Zachary Levonian More data munging; still issues identifying root journal and guestbooks
2022-02-17 7a2d2ef Ngan Nguyen move back to notebook folder
2022-02-17 c178c89 Ngan Nguyen move back to notebook folder
2022-02-17 15b29f1 Ngan Nguyen Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2022-02-17 60e321d Ngan Nguyen evaluate learning rate
2022-02-15 4ad7c9b Zachary Levonian Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2022-02-15 22e4ed1 Zachary Levonian Checkpoint with broken activity extraction implementation
2022-02-14 4e5e846 Ngan Nguyen Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2022-02-10 8be7be5 Matthew formatting fix
2022-02-10 3db3bfd Matthew adding adam optimizer experiment
2022-02-10 043b477 Ngan Nguyen Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2022-02-08 89cc12d Zachary Levonian Added code to extract subset of relevant data for relevant users and sites
2022-02-04 0d3d279 Zachary Levonian Added default Adam parameters, and config parameter to turn off Learning Rate scheduler.
2022-02-04 1f44f78 Zachary Levonian Fxied preprocess script by enforcing int type, and removing to correct indentation level.
2022-02-03 9dff4e0 Ngan Nguyen Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2022-01-31 55d478a Zachary Levonian Update config to generate eval script with more realistic total time, and added additional example code to eval notebook, including an analysis demonstrating that MRR does not decrease over time and generally is invariant wrt week of 2021.
2022-01-30 bd9ddca Ngan Nguyen Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2022-01-30 9e3d04b Ngan Nguyen experimenting evaluation on models
2022-01-28 3f89895 Zachary Levonian Added sample notebook demonstrating a few approaches to evaluation, fixed preprocessing bug when the FeatureManager is reinstantiated, and updated evaluation script to support multiple models by default.
2022-01-26 5132eea Zachary Levonian Added temporary script to test model evaluation.
2022-01-26 91eebd0 Zachary Levonian Updated evaluateModels script to produce the test metadata with model metrics for the given scorer. It's not possible to fully evaluate a model.
2022-01-26 e0f3773 Zachary Levonian Rewrite column dropping in preprocess, including adding the ability to specify glob patterns and an extension of get_feature_indices, update ModelTrainer and the SimNet implementation to use feature info from the FeatureManager, add config for LinearNet feature ablations, refreshed training data matrices based on recent snapshot.
2022-01-25 3852c43 Zachary Levonian This commit adds a new evaluation flow (evaluateModels.py), a new text-feature cache and pooler, a new no-parameter cosine similarity model, and adds structure to cbrec.experiment's generation. A variety of bugs and documentation points have been fixed or added.
2022-01-24 a322a77 Zachary Levonian Updated text feature extraction for training data
2022-01-21 eb54799 Zachary Levonian Rerunning full feature extraction process with Jan 20, 2022 data dump
2021-12-28 73b48ec Zachary Levonian Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2021-12-28 5ee1bef Zachary Levonian Fix to v2 annotation spreadsheet
2021-12-23 fd65f8a Ngan Nguyen Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2021-12-23 011f3fe Ngan Nguyen added concatnate and max pool for text features
2021-12-22 7b41956 Zachary Levonian Added v2 click data annotations. Updated spreadsheet. Added notes on potential third, random sample of journal updates for annotation.
2021-12-16 8496f3a Zachary Levonian Finished analysis of comments timestamp issue.
2021-12-16 fe93eb7 Zachary Levonian Added batch experiment analysis notebook demonstrating loading a batch of experiments.
2021-12-15 b287501 Zachary Levonian Reconsidering default value
2021-12-14 f53f0a0 Zachary Levonian Simple patch to set a reasonable default for time_to_most_recent features.
2021-12-10 a0faf84 Matthew removing excess tests
2021-12-10 a14f687 matthewmac99 Delete slurm-8898607.out
2021-12-10 46d92f9 Matthew Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2021-12-10 06ac605 Matthew working experiment code
2021-12-05 f0429c5 Zachary Levonian Refreshed cloudfront data. Added better data handling for first clicks and v1 sampling of journal annotations.
2021-12-03 cd9a8ec Zachary Levonian Copy activity monitoring notebook
2021-12-03 1872f25 Zachary Levonian Sent batch_id 11 qualitative feedback requests.
2021-12-02 931e6b0 Zachary Levonian Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2021-12-02 01c7733 Zachary Levonian Updated activity monitoring notebook with some hacky code to generate qualitative email contents
2021-12-02 2c72153 zentavious k-buckets
2021-11-30 345ad8a Ngan Nguyen added different pools for features
2021-11-30 dbea2b2 Matthew test config generation is functional
2021-11-24 4d04d5a Zachary Levonian batch_id 10 sent.
2021-11-19 97d9d87 Zachary Levonian Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2021-11-19 8e320d8 Zachary Levonian Updated with guestbook analysis as well.
2021-11-19 c2660d6 matthewmac99 Update model_gen_script.sh
2021-11-19 f5691d2 matthewmac99 Update model_gen.py
2021-11-19 faf4e3e matthewmac99 Update model_gen.py
2021-11-19 ececbdd matthewmac99 Update model_gen_script.sh
2021-11-19 adc86ae matthewmac99 Update model_gen_script.sh
2021-11-18 4b0fdcc Zachary Levonian Added investigation of timestamp bug
2021-11-17 b24f017 zentavious analysis of duration features, bug fix for duration bins
2021-11-17 c28580c Zachary Levonian batch_id 9 complete.  Small updates to feature extraction to make computation faster.
2021-11-16 3b028fc Matthew input folder
2021-11-15 e50b5ab Matthew scripts are now compatible with jsons
2021-11-11 0b94181 matthewmac99 Update model_gen_script.sh
2021-11-11 c91881c matthewmac99 Updated to reflect changes in Python script
2021-11-11 426ffa1 matthewmac99 Updated script with eval()
2021-11-10 5fa0546 Zachary Levonian batch_id 8 complete.
2021-11-08 72afca7 Matthew removed print statements
2021-11-08 f9fdb79 Matthew initial experiment scripts
2021-11-08 6a24aee Matthew McNamara removed '=' sign from inside brackets on lines 35, 38 to stop compilation error
2021-11-04 ba4ae63 zentavious Merge branch 'main' of https://github.com/umncs-caringbridge/recsys-peer-match into main
2021-11-04 f6bea12 zentavious experiment/ minor bug fix in preprocessor
2021-11-04 08cb946 Zachary Levonian Updated activity monitoring notebook with analysis pre-meeting
2021-11-03 348eba9 Zachary Levonian Updated activity monitoring notebook with counts for visit, follow, and interaction actions.
2021-11-03 d37a9c8 zentavious * bug
2021-11-03 9afc56b zentavious Merge branch 'main' of https://github.com/umncs-caringbridge/recsys-peer-match into main
2021-11-03 e7604ee zentavious adding feature_keys to memory in feature_manager
2021-11-02 83e53f6 Zachary Levonian batch_id 7 complete. Added function to make it easier to reload model checkpoint.
2021-11-02 8e194d4 Zachary Levonian Updates for batch 7
2021-11-02 98c8cee Zachary Levonian Added stub code demonstrating approach to building a feature index map
2021-11-02 7a0c4d7 zentavious all durations are hours
2021-11-01 5268534 Zachary Levonian Updated triple generation to enable interactive mode more easily, additional debugging of misaligned candidate USPs.
2021-11-01 c10cdec Zachary Levonian Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main. May have accidentally overwritten preprocess_encode_columns"
2021-11-01 bac5d66 Zachary Levonian Bug fixes and updates. Investigating erroneously eligible USPs in test reccontexts.
2021-11-01 d4273a4 zentavious bug fixes for preprocessing and private methods
2021-10-28 49e98a6 zentavious Merge branch 'main' of https://github.com/umncs-caringbridge/recsys-peer-match into main
2021-10-28 8da66f1 zentavious preprocessing, removal/hot encoding
2021-10-28 b1611e6 Zachary Levonian Checked out the weight-decay results in a new notebook, and updated text feature creation script to optionally override the output directory.
2021-10-27 322c88f Zachary Levonian Starting second weight-decay experiment
2021-10-27 1eef6d6 Zachary Levonian Added temporary sbatch script to test out a very long model training period.
2021-10-27 bcb565b Zachary Levonian Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2021-10-27 5b0fe6c Zachary Levonian Batch_id 6 sent out.  Added some functionality to train with weight decay, added templates, added survey links.
2021-10-27 2b27b49 matthewmac99 Changing Mail Address to "xxx@umn.edu" for testing purposes.
2021-10-26 1a517f0 Zachary Levonian Updated notebooks for new batch of data, documentation updates and initial experiments with model_manager
2021-10-26 fb7fd2f Zachary Levonian Bug fixes and feature additions
2021-10-26 7b509ea Zachary Levonian Added and tested a save/load implementation for the model weights and the training metrics. Documentation updates. Added /modeling demo in experiments notebook.
2021-10-25 eeab74d Zachary Levonian Various bug and code-flow fixes, and an untested implementation of model saving and loading.
2021-10-25 78aaaf9 Zachary Levonian Added notebook to produce codes for Miro affinity mapping. Small update to torch experiments documentation
2021-10-23 cd22064 zentavious syntax errors, initial tests of manager train pipeline
2021-10-21 7f62dcd Zachary Levonian Added initial modeling subpackage that breaks out the functionality of TorchModel into separate components that can be controlled via configuration. Absolutely none of this code is tested, so it still needs to be massaged and tested.  Key design decisions to make yet about what, how, and when to save after training a model.
2021-10-20 a02671c Zachary Levonian Fixed bug where target_inds are computed incorrectly after resorting.
2021-10-20 193c6d2 Zachary Levonian Batch_id 5 (Batch 6) sent. Miscellaneous bugfixes and documentation updates.
2021-10-18 2b2fee8 Matthew McNamara training and evaluation on personal Jupyter notebook
2021-10-17 e6b9d06 Matthew McNamara Merge branch 'main' of https://github.com/umncs-caringbridge/recsys-peer-match into main
2021-10-15 d65854f zentavious experiementing with scoring, syntax fix
2021-10-12 a1619b9 Zachary Levonian Update Zach's notebook with scorer example from today's meeting.  Small bugfixes, plus new function in feature_loader.
2021-10-12 03728f7 Zachary Levonian Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main.
2021-10-12 b8eae79 Zachary Levonian Batch_id 4 updates.  Updated prediction script to auto-save the required journal ids, added script to ensure those journal ids are present in the text database, and partially updated templates.  Added new survey links.
2021-10-11 16d6a3d zentavious Merge pull request #2 from umncs-caringbridge/zentx005/test_accuracy
2021-10-11 241313a zentavious nit
2021-10-11 03ac307 zentavious saving metrics and plots, optional display
2021-10-08 0817a85 zentavious Revert epoch change
2021-10-07 25ab21b zentavious updating torch model/training logic to log validation loss
2021-10-07 a4c0928 zentavious Merge branch 'main' of https://github.com/umncs-caringbridge/recsys-peer-match into zentx005/test_accuracy
2021-10-07 f41e96c zentavious testing changes for plotting validation acc/loss
2021-10-07 7ceea7e Zachary Levonian Added notebook to create data for annotation
2021-10-07 c6d28e7 Zachary Levonian Update documentation and dates in script
2021-10-06 3295446 Zachary Levonian Documentation updates and bugfix to candidate feature mean pooling.
2021-10-05 0df9a55 Zachary Levonian Updated documentation
2021-10-05 e889488 zentavious Merge branch 'main' of https://github.com/umncs-caringbridge/recsys-peer-match into zentx005/test_accuracy
2021-10-04 ba2e20c Ngan Nguyen Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2021-10-04 957e655 Ngan Nguyen added my notebook
2021-10-04 fe06a44 Matthew McNamara working through basic neural network model
2021-10-04 e19664e zentavious Log accuracy against test set
2021-10-04 41b6896 zentavious Merge pull request #1 from umncs-caringbridge/master
2021-10-04 53f37e0 Matthew Zent Revert "Implimented testing accuracy on test set"
2021-10-04 b9eb4fb Matthew Zent Implimented testing accuracy on test set
2021-10-04 1401439 Matthew Zent playing with logging test accuracy instead of training accuracy
2021-10-01 35bb908 Zachary Levonian Add notebooks at end of b3 (batch 4).
2021-10-01 258b143 Zachary Levonian Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2021-10-01 498ff04 Zachary Levonian updated feature loader, updated notebook to include an example of model training using cached feature arrays
2021-10-01 562ebf3 Matthew McNamara adding my TorchExperiment file
2021-09-30 6cf517f Zachary Levonian Updated prediction script to infer a time 6 hours beyond the last interaction to decrease the likelihood of temporal proximity effects, updated feature loader to further replace the functionality of torchmodel, updated torchmodel's scoring interface for test data to use a new RecContext.X_test key
2021-09-30 844b177 Zachary Levonian Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2021-09-30 853d320 Zachary Levonian Added hacky script to train model
2021-09-29 b8765dc Matthew Zent Added zent_torch_experiments notebook as follow-up to meeting
2021-09-28 7b15ad9 Zachary Levonian Added torch experiments notebook during meeting
2021-09-27 6c18fff Zachary Levonian Added stub for a feature loader class that manages database connections to feature data independent of training and prediction.
2021-09-24 06cd061 Zachary Levonian Updates to notebooks after batch 3 sent out.
2021-09-24 2559584 Zachary Levonian Bug fixes and updates to training process
2021-09-24 60f1340 Zachary Levonian Added updated cloudfront data and survey links for third batch
2021-09-23 798cd65 Zachary Levonian Added script to top up test2train text features.
2021-09-17 ec19ce9 Zachary Levonian Updated notebooks with code that produced the emails for batch 1
2021-09-17 2c500c9 Zachary Levonian Refreshed with fixed journal data
2021-09-17 6114ebb Zachary Levonian Added code to enable using test data to train the model
2021-09-17 3e9752b Zachary Levonian Added links to b1 survey feedback
2021-09-17 4e3b4e5 Zachary Levonian Added separate template file, and added batch 1 templates.
2021-09-15 fd1be86 Zachary Levonian Added additional testing and test2train pipeline.
2021-09-15 4caf670 Zachary Levonian Candidate identification with new data dump.
2021-09-15 d5e6b5c Zachary Levonian Bug fixes, including the addition of command-line flags to skip feature generation, additional documentation, and a new test2train function in torch_model.
2021-09-15 494f666 Zachary Levonian Update with new data dump, added additional activity monitoring data
2021-09-15 9d7c0af Zachary Levonian Added feedback data file
2021-09-15 2c40b86 Zachary Levonian Added script to add all required texts for prediction at once.
2021-09-09 0f19396 Zachary Levonian Fixed bug in amp noising, restarted job.
2021-09-09 ee60890 Zachary Levonian Updated PytorchTraining notebook to save as many details as possible as JSON. Updated cbrec to use published_at rather than created_at for journals, and to add noise to amps.
2021-09-09 14ed1db Zachary Levonian Updated cloudfront click data, and added notebook to fix original amp timestamps based on a sampling process from the new reactions.
2021-09-03 33d258d Zachary Levonian Added initial attempt at activity monitoring.
2021-09-02 77eafa6 Zachary Levonian Updated prediction helper notebook to identify journal ids that will be needed at prediction time.
2021-09-02 1d2c873 Zachary Levonian Wrote out participant ids to a file, along with other study info collected both from the survey and manually.
2021-09-02 b9abe68 Zachary Levonian Rewrote compose functionality to template the plaintext and HTML formats of the rec emails.  Added untested functionality to record whenever an email is sent.
2021-09-02 9ea183e Zachary Levonian Add in manual opt-ins.
2021-09-02 63a697a Zachary Levonian Update to recent data
2021-09-02 5595c19 Zachary Levonian Updated with code for making predictions
2021-09-02 b99f25d Zachary Levonian Update to time req and docs
2021-09-02 0a4d757 Zachary Levonian Added new participant secret value tsv
2021-09-02 219b6b9 Zachary Levonian Bugfixes and added cache and allowlist to decrease duplicate effort
2021-09-01 275f236 Zachary Levonian Add distribution history / personalized links for feedback survey
2021-08-31 a1ad191 Zachary Levonian A variety of feature additions and bugfixes, including changes to the linear torchmodel learning rate, helper functions for loading checkpoints and the like, and additional command-line flags during prediction.
2021-08-31 734e8ec Zachary Levonian Wrote eligible participant ids to a file for prediction
2021-08-31 2bb685a Zachary Levonian Update training notebook to produce model that incorporates text features, add notebook for computing the journal oids of active, eligible users at a checkpoint.
2021-08-31 0d84dba Zachary Levonian Added helper scripts for producing texts for prediction.
2021-08-30 f4db3a7 Zachary Levonian A few bugfixes, updated TorchModel to use text features, added script to add missing train journal features, added utility functions for retrieving text features in embeddingdb, added class for identifying recent journal updates in journalid (using bisect).
2021-08-29 c7d1052 Zachary Levonian Added notebook to identify site profiles of users who are not participants.
2021-08-29 e7c8283 Zachary Levonian Refresh with new data snapshot
2021-08-29 e1b91c6 Zachary Levonian Added matched data outputs, added new survey data
2021-08-29 2b7c207 Zachary Levonian Add script for continuation
2021-08-29 6e3d684 Zachary Levonian Config updates, with some convenience functions added.
2021-08-29 75f5a6a Zachary Levonian Allowed population of journal_dict from updated journal data
2021-08-29 2143b8e Zachary Levonian Minor updates to logging
2021-08-29 d64635a Zachary Levonian Began updating prediction, added additional CLI options to the triple_generation script for debugging pickle issues.
2021-08-28 f89b4ab Zachary Levonian Updated rec process to be more sensible about checkpoints, to respect existing metadata and features generation in the database, and to log with greater detail in some places.
2021-08-22 bf6a681 Zachary Levonian Additional exploration of participant data
2021-08-22 c0f22c0 Zachary Levonian Reran with new data, and added initial analysis of participants
2021-08-22 d2db1c3 Zachary Levonian Added analysis of new reactions
2021-08-20 3969ed8 Zachary Levonian Added matched email list for new dump
2021-08-19 2234571 Zachary Levonian Data update with latest respondents
2021-08-18 e2c4a98 Zachary Levonian Updated number of cores
2021-08-18 c1d8991 Zachary Levonian Added notebook to identify journal oids from the metadata produced by gen, and small updates to text extraction code.
2021-08-18 1b97b6b Zachary Levonian Updated script to use roberta for extraction. Added shell script to run as qsub
2021-08-18 c4c7f86 Zachary Levonian Created text module within cbrec, including version of threading process that works via pool.
2021-08-17 7720408 Zachary Levonian Updated generation process to NOT record journal ids for candidate usps, for space reasons.
2021-08-17 7674685 Zachary Levonian Decrease memory requirement after fixing bug
2021-08-17 bef2c99 Zachary Levonian Fixed use of non-string key name in metadata dict
2021-08-16 4a16b93 Zachary Levonian Add further memory cap, while trying to debug where the memory explosion is coming from.
2021-08-16 c31beca Zachary Levonian Updated cbrec to allow greater memory usage and to free journal data after it is transformed into a time-aware dict.
2021-08-16 138ad4a Zachary Levonian Refresh with new survey data.
2021-08-07 8a84462 Zachary Levonian Updated with additional computation time, so that the job is more likely to complete in one go.
2021-08-07 5e5cf2d Zachary Levonian Updated reccontext to include the jouranl oids of the source, target, and candidates, so they can be used post-processing for extracting text features.
2021-08-05 33525e8 Zachary Levonian Refreshed survey notebook and survey data. Added code to compute per-response counts for the multi-choice Qs.
2021-08-04 7a642bc Zachary Levonian Changes to feature extraction process for texts.
2021-08-04 86261e6 Zachary Levonian Added initial survey dump and data analysis of survey dump.
2021-08-04 0e0fdad Zachary Levonian Updated data paths and filenames. Added initial text implementation with helper functions for accessing data.
2021-08-03 3ef071e Zachary Levonian Updated sender alias to be study-specific.
2021-07-29 3a43504 Zachary Levonian Added initial script for sending email from my UMN account.
2021-07-27 72ecda1 Zachary Levonian Added comparison of journals in new and 2019 dumps. Updated site duration analysis using new data
2021-07-27 b6a51f6 Zachary Levonian Updated interaction extraction to work with new data and update valid author selection
2021-07-27 3178f01 Zachary Levonian Updated to work with new data source
2021-07-22 72793d7 Zachary Levonian Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2021-07-22 0a55030 Zachary Levonian Added design files
2021-07-19 6c049e6 Zachary Levonian Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2021-07-19 2b5e838 Zachary Levonian Added analysis of site descriptions using newest data. Renamed notebook.
2021-07-04 59a6b3e Zachary Levonian Update README with additional resources referenced
2021-07-02 9a79dca Zachary Levonian Add README with other BPR implementations
2021-06-28 aa3265f Zachary Levonian Finished site survival time analysis.
2021-06-28 4afa174 Zachary Levonian Updated notebook with additional analyses
2021-06-25 c4b77ca Zachary Levonian Added initial version of notebook for reporting high-level details about site survival time.
2021-06-09 7d109cb Zachary Levonian Updated survival analysis notebook with Kaplan Meier survival time estimates, verifying expected patterns. Quant slide summarizing results added to slide deck.
2021-06-09 5c527ac Zachary Levonian Added script for precomputing early site interaction counts at some threshold.  Updated script to build a site_df suitable for survival modeling
2021-06-08 d1cecb5 Zachary Levonian Added initial notebook for an analysis to determine if early interactions from authors are different from early interactions by others. Seems like the interaction processing code may be a little slow, although it may have just been the network connection.
2021-06-07 aca7552 Zachary Levonian Updated packages available in pytorch-cpuonly
2021-05-28 5534976 Zachary Levonian Updated script to give even more time and memory
2021-05-28 0b397de Zachary Levonian Added torchmodel module for defining and training PyTorch based recommendation models. Initial fully-connected network does better than any of the baselines on all metrics.
2021-05-28 f55172c Zachary Levonian Updated readme info
2021-05-28 7400dc1 Zachary Levonian Added notebook to train sklearn model from real data, and added stub notebook for text feature extraction. New MSI conda env: pytorch-cpuonly, for doing roberta feature extraction.
2021-05-27 2a227ce Zachary Levonian Fixed featuredb multi-type support, updated db test script to test featuredb multi-type support, added sklearn model scorer to feature extraction, added script that demonstrates loading data and training a model.
2021-05-27 89416e2 Zachary Levonian Updated featuredb to store numpy datatype in an extra byte.
2021-05-27 89b166c Zachary Levonian Updated featuredb to retrieve appropriate arrays from test_contexts, added untested sklearn model scorer, moved baseline scorer functionality to parent class, other code and bug fixes. Added utils for external notebook use.
2021-05-26 3fb99f3 Zachary Levonian Added entry point for prediction, using a cached checkpoing rdg. Did some reorganizing, updated the creation of targetless rec contexts. Added stub notebook for debugging training from the feature db.
2021-05-26 8f9bf30 Zachary Levonian Added prediction period after the test period, wherein a sample of recs are generated. Bug fixes and performance optimizations in several inner methods.
2021-05-25 494f7bd Zachary Levonian Updated coverage to limit memory usage for specific sites that are frequently ranked.
2021-05-24 f73e4f9 Zachary Levonian Adjusted time to run based on previous execution
2021-05-24 6efc304 Zachary Levonian Assorted bugfixes, and added coverage module.  Added stub doctest file. Added initial implementation of coverage that dumps coverage stats in a pickled dictionary.
2021-05-21 dfcbedf Zachary Levonian Added initial CoverageTracker implementation. Updates to config and eval to support tracking of train period initiations for better test performance disambiguation.
2021-05-20 b021936 Zachary Levonian Added mechanism for controlling commit rate to database, along with additional debugging details and config options.
2021-05-19 0199c2a Zachary Levonian Fixed eval bug and added notebook to explore Baseline results from the full testing period.
2021-05-19 4c2bca3 Zachary Levonian Attempt to bugfix issue with ndcg score on arrays containing inf
2021-05-18 63bf736 Zachary Levonian Small update to slurm config
2021-05-18 58e7f6a Zachary Levonian Bugfix to config
2021-05-18 662b6a1 Zachary Levonian Updated cbrec to allow checkpointing using pickle.
2021-05-18 33608d2 Zachary Levonian Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2021-05-17 dce3e39 Zachary Levonian updated slurm script based on initial performance profiling
2021-05-17 ae8c0b0 Zachary Levonian Bug fixes, added notebook for analyzing baseline data.
2021-05-14 c1644f7 Zachary Levonian Bug fixes. Metric generation still known to be broken.
2021-05-13 32fa32a Zachary Levonian Updated RecContext workflow by separating database writing and feature generation. Need to refine the state set in RecContext that defines whether it is eligible or not for feature generation + baselines.
2021-05-12 571f84f Zachary Levonian Added CosineSimilarity baseline and fixed cohort baseline. Restructured classes to better have single responsibilities.
2021-05-11 4757c41 Zachary Levonian Updated TestConfig to use testing paths and to generate data to test output paths. Began constructing RecContext in triple generation, although the flow needs to be tweaked.
2021-05-11 4fd3f2d Zachary Levonian Added test data.
2021-05-11 0bde636 Zachary Levonian Began implementing test workflow. Added script to generate test data.
2021-05-10 ba59511 Zachary Levonian Added baseline implementations to evaluation, along with a flow for building feature matrices, scoring them, and computing metrics from those scores.
2021-05-07 d62bf02 Zachary Levonian Updated eval to implement the two-stage merge and reduce process.
2021-05-07 836583a Zachary Levonian More progress updating the extraction process.
2021-05-05 d2a100f Zachary Levonian Revamping the way that training & test rec contexts are generated and the way metadata is tracked.
2021-05-04 13041b3 Zachary Levonian Began making changes to actually compute baselines, but there's still a ton to do.  One specific note: I still need to generate counts for EXISTING vs ELIGIBLE authors when negative sampling, in order to determine coverage. Probably only need to do this during test generation.
2021-05-04 ab1c75b Zachary Levonian Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2021-05-04 76d5579 Zachary Levonian Changed selection criteria
2021-04-23 5256f56 nguy4068 fix bug
2021-04-22 640ce09 Zachary Levonian Began updating triple generation to avoid dependence on user_site_df and to process journal updates to determine eligibility.
2021-04-20 56d99bd Zachary Levonian Added shell script for executing gen via sbatch, updated recent activity counter to produce a sorted list of all active users, and fixed a few bugs in feature_extraction
2021-04-16 9e40b9b Zachary Levonian updates to generation process to advance to the point where features are actually generated from the graph and recent activity.
2021-04-16 42a7123 Zachary Levonian Small update to scratch code.
2021-04-15 2f88836 nguy4068 update viewer site
2021-04-12 3a92794 Zachary Levonian Added implementation of feature extraction. Added code for network and activity feature computation. Still need to actually compute features into vectors and save them to the database.
2021-04-08 a726a6c Zachary Levonian Terminated initial run for debugging. A problem with the ordering of the journal_dict, otherwise looking good.
2021-04-08 b240cd4 Zachary Levonian Added stub core loop for generating training and testing triples.
2021-04-08 3398b2b Zachary Levonian Small updates to config, network, and triple_generation code.
2021-04-06 ac2c061 Zachary Levonian Added further sketch notes on implementation. Added stub helper files for actually generating triples and for tracking the network
2021-04-05 62ca0fb Zachary Levonian Cleaned up valid author selection, finished initial pass on interaction extraction.
2021-04-05 a5a6ed2 Zachary Levonian Added stub guestbook extraction notebook, although it was replaced with scripts in caringbridge_core. Added stub take interaction extraction with more int types
2021-04-05 991fa5e Zachary Levonian Updated author selection criteria and added some analysis of 'extreme' user/site pairs.
2021-04-04 6f50827 Zachary Levonian Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2021-04-04 ede7c3c Zachary Levonian Added initial attempt at infrastructure for a feature database to support generation of training and testing triples. Decided on minimal state in the database. Current implementation makes no attempt to distinguish train from test.
2021-04-02 dafb4e1 Zachary Levonian Added scrolling functionality to Comments
2021-04-02 a1e2785 nguy4068 change
2021-04-01 5e0fb00 Zachary Levonian Added stub notebooks for identifying valid authors and extracting necessary interactions
2021-04-01 806c051 Zachary Levonian Rearranging some files
2021-03-29 b147795 Zachary Levonian Fixed kernel spec.  Updated test data to fix bug in site 2 journals and to add some additional activity.
2021-03-19 aab2693 Zachary Levonian Synthetic BPR documentation updated
2021-03-19 a03f01c Zachary Levonian Added stub implementation of backend access to user data
2021-03-19 cf827ac nguy4068 change
2021-03-15 bd596bd Zachary Levonian Added stub synthetic bpr notebook
2021-03-11 722f6ab nguy4068 Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2021-03-06 5ef3af8 nguy4068 change journal
2021-03-06 ede2916 Zachary Levonian Updated data utils to ensure an empty array is passed from the backend when no amps or comments are available.
2021-03-05 cc7a92c Zachary Levonian Removing Mac OS X files from version control.
2021-03-04 2ffd31d nguy4068 changing siteViews adding Journal and GuestBook components
2021-03-01 7979e14 Zachary Levonian Update to notebook documentation.
2021-03-01 5e3d3ba Zachary Levonian Updated SiteView to load the site data via fetch() when the page is loaded.
2021-03-01 092a07e Zachary Levonian Added new site data endpoint, and added resources for database usage. Data utils uses the databases to structure the data in the prod or test databases. Configs define the paths needed to load resources appropriately.
2021-03-01 58e0a4e Zachary Levonian Updated startup to use explicit app factory, and added switch for dev vs MSI config.
2021-03-01 e4619a6 Zachary Levonian Fixed mistakenly empty journal text test data.
2021-03-01 06820cb Zachary Levonian Further updates to test data.
2021-03-01 aaacd32 Zachary Levonian Fixed data
2021-03-01 eb3f15d Zachary Levonian Added test databases.
2021-03-01 c88d9df Zachary Levonian Added notebook to create journal metadata and amps databases. Added notebook to create test data.
2021-03-01 a653768 Zachary Levonian Removed test file
2021-03-01 91f85dc Zachary Levonian Added DS_Store to git exclusions
2021-03-01 5411862 Zachary Levonian Added hooks for DB creation and for differing configs on MSI
2021-02-28 16ed7be Zachary Levonian Refactored names to better reflect js naming conventions. Also added back button to site view.
2021-02-28 fb85512 Zachary Levonian Fixed routes to properly populate the siteId
2021-02-26 8fb4e93 nguy4068 change version of react router back to 6.0.0
2021-02-25 1641600 nguy4068 update new route for SiteView component, create DetailView
2021-02-12 0488d43 nguy4068 Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2021-02-11 bf7c8c3 Hai Ngan Nguyen changes index.py SiteView and ResultViewer
2021-02-05 5a87759 Zachary Levonian Merge branch 'main' of github.com:umncs-caringbridge/recsys-peer-match into main
2021-02-05 6aa5ef0 Zachary Levonian CHanged up a few of the descriptions
2021-01-25 b8fbe65 Zachary Levonian Updated with additional links.
2021-01-25 30c941e Zachary Levonian Added front-end implementation from Devias IO example code.
2021-01-25 95348e6 Zachary Levonian Updated respository with stub backend.
2021-01-22 ebce253 Ngan Nguyen First Commit
2021-01-22 dfb8540 Ngan Nguyen Updated readme
2021-01-22 4512e3b Zachary Levonian Initial commit