| Age | Commit message (Collapse) | Author | Lines | |
|---|---|---|---|---|
| 2024-01-25 | Use LMT algorithm defaults for max depth and min samples leaf. | Birte Kristina Friesel | -3/+3 | |
| max bins remains at 120 (≠25), but that one should only affect run-time and not accuracy/complexity | ||||
| 2024-01-25 | XGB: switch to XGBoost defaults | Birte Kristina Friesel | -9/+68 | |
| 2024-01-25 | add an optional XV progress bar | Birte Kristina Friesel | -0/+19 | |
| 2024-01-25 | LMT: Tailor hyper-parameters towards higher accuracy (and longer training time) | Birte Kristina Friesel | -1/+54 | |
| Also, allow users to override them | ||||
| 2024-01-25 | add hyper-parameter export for CART and LMT | Birte Kristina Friesel | -0/+20 | |
| 2024-01-24 | Use separate /hyper/ hierarchy for hyper-parameter export | Birte Kristina Friesel | -13/+20 | |
| 2024-01-23 | --function-override: Use : for splitting to support spaces in key/attr names | Birte Kristina Friesel | -1/+1 | |
| Also, document it in doc/modeling-methods | ||||
| 2024-01-23 | Allow --skip-param-stats to be used in FIT_FOL mode | Birte Kristina Friesel | -5/+22 | |
| 2024-01-22 | Support LUT model independent of --skip-param-stats | Birte Kristina Friesel | -6/+7 | |
| 2024-01-22 | Add --skip-param-stats support to analyze-{archive,log} | Birte Kristina Friesel | -0/+8 | |
| 2024-01-19 | RMT: Set DFATOOL_RMT_WEIGHTED_AVG=1 for weighted average in queries | Birte Kristina Friesel | -0/+6 | |
| 2024-01-19 | Store n_samples in all relevant ModelFunction instances | Birte Kristina Friesel | -7/+14 | |
| 2024-01-19 | functions: explicitly mention _xgb_ hyper-parameters in dref export | Birte Kristina Friesel | -7/+7 | |
| 2024-01-19 | dataref export: include XGB hyper-parameters | Birte Kristina Friesel | -1/+15 | |
| 2024-01-19 | optionally store n_samples in ModelFunction | Birte Kristina Friesel | -17/+29 | |
| 2024-01-19 | Remove unused ModelAttribute.split code; splits are handled via SplitFunction | Birte Kristina Friesel | -80/+34 | |
| 2024-01-19 | parameters: nits | Birte Kristina Friesel | -4/+6 | |
| 2024-01-19 | assertions | Birte Kristina Friesel | -1/+4 | |
| 2024-01-18 | Rename DFATOOL_{'' → ULS_}SKIP_CODEPENDENT_CHECK | Birte Kristina Friesel | -1/+1 | |
| 2024-01-18 | kconfig: ignore strings by default (implicit DFATOOL_KCONF_IGNORE_STRING=1) | Birte Kristina Friesel | -1/+1 | |
| 2024-01-18 | build_dtree: do not ignore codependent parameters | Birte Kristina Friesel | -2/+7 | |
| Ignore non-numeric function arguments instead. Strings arguments are often paired with a length indicater and we are only interested in the latter. | ||||
| 2024-01-18 | parameters: ModelAttribute: remane .ignore_param to .ignore_codependent_param | Birte Kristina Friesel | -12/+14 | |
| This should help avoid confusion. | ||||
| 2024-01-17 | paramfit: use sum(ssr) rather than mean(rmsd) for best-fit selection | Birte Kristina Friesel | -9/+21 | |
| 2024-01-17 | cli: fix energy_Pt model check | Birte Kristina Friesel | -2/+2 | |
| 2024-01-17 | _try_fits: remove unused results_by_param | Birte Kristina Friesel | -10/+0 | |
| 2024-01-17 | Disable DFATOOL_DTREE_IGNORE_IRRELEVANT_PARAMS by default | Birte Kristina Friesel | -2/+2 | |
| It is not working properly in 95% of cases | ||||
| 2024-01-17 | dtree irrelevant param detection nits; still not working properly | Birte Kristina Friesel | -5/+5 | |
| 2024-01-17 | Stop splitting if there is no split that reduces loss | Birte Kristina Friesel | -1/+3 | |
| 2024-01-16 | build_dtree: add another assertion for good measure | Birte Kristina Friesel | -4/+2 | |
| 2024-01-16 | model: Correctly set threshold from compute_stats | Birte Kristina Friesel | -1/+1 | |
| 2024-01-16 | model: ModelAttribute is defined in .parameters, not .automata | Birte Kristina Friesel | -1/+2 | |
| 2024-01-16 | Add LMT support for --export-json / ModelAttribute.to_json | Birte Kristina Friesel | -1/+5 | |
| 2024-01-16 | plotter: remove useless 'red' colour specifier | Birte Kristina Friesel | -1/+1 | |
| 2024-01-16 | Implement --show-model=param for LMT | Birte Kristina Friesel | -0/+55 | |
| 2024-01-15 | improve(?) loss_ignore_scalar implementation: more opportunities for FFS | Birte Kristina Friesel | -6/+44 | |
| 2024-01-15 | dtraa loss_ignore_scalar: calculate unsuitable params for each node | Birte Kristina Friesel | -2/+20 | |
| The global list may no longer be helpful after the observations have been partitioned by parent nodes | ||||
| 2024-01-15 | dtree: replace condition by assertion | Birte Kristina Friesel | -11/+11 | |
| 2024-01-15 | decrease non-debug verbosity | Birte Kristina Friesel | -6/+2 | |
| 2024-01-15 | PTAModel.asses: support transitions with partial data | Birte Kristina Friesel | -1/+1 | |
| This can happen when --filter-observation is in use | ||||
| 2024-01-12 | add --filter-observations | Birte Kristina Friesel | -0/+23 | |
| 2024-01-12 | Move --force-tree definition to cli.py | Birte Kristina Friesel | -0/+5 | |
| 2024-01-12 | --export-dref: store commandline in .tex file | Birte Kristina Friesel | -0/+2 | |
| 2024-01-12 | Expose DFATOOL_ULS_MIN_DISTINCT_VALUES training hyper-parameter | Birte Kristina Friesel | -1/+5 | |
| 2024-01-12 | It's THRESHOLD, not TRESHOLD. Oops. | Birte Kristina Friesel | -2/+2 | |
| 2024-01-11 | Rename --show-model-size to --show-model-complexity | Birte Kristina Friesel | -17/+18 | |
| also: make it work for non-tree models | ||||
| 2024-01-11 | Rename --show-quality=table to --show-model-error | Birte Kristina Friesel | -0/+5 | |
| 2024-01-11 | move --show-quality=table header to cli.py | Birte Kristina Friesel | -1/+8 | |
| 2024-01-11 | add --export-json option | Birte Kristina Friesel | -0/+6 | |
| 2024-01-11 | move --plot-param definition to cli.py | Birte Kristina Friesel | -0/+9 | |
| 2024-01-11 | I need more coffee | Birte Kristina Friesel | -0/+18 | |
 |
index : dfatool | |
| Performance Model Generation for Configurable System Components | derf |
| summaryrefslogtreecommitdiff |