Age | Commit message (Collapse) | Author | Lines | |
---|---|---|---|---|
2021-03-16 | move codependent parameter detection to Model / ModelAttribute | Daniel Friesel | -1/+4 | |
Still TODO: Ignore codependent parameters when partitioning data for analytic modeling / regression | ||||
2021-03-08 | model.assess: remove superfluous "by_name" indirection | Daniel Friesel | -11/+8 | |
2021-03-08 | add dataref export | Daniel Friesel | -0/+42 | |
2021-03-05 | promote sub-state models to a first-class modeling citizen | Daniel Friesel | -64/+23 | |
get_fitted_sub is no longer present, sub-state models are part of get_fitted now | ||||
2021-03-04 | Restore --export-energymodel | Daniel Friesel | -2/+1 | |
2021-03-03 | Remove ModelInfo; add info to ModelFunction instead | Daniel Friesel | -4/+4 | |
2021-03-02 | ModelAttribute + sub-classes: add to_json method | Daniel Friesel | -3/+14 | |
2021-03-02 | PTAModel: Turn states, transitions, and states_and_transitions into attributes | Daniel Friesel | -33/+24 | |
2021-03-02 | AnalyticModel, PTAModel: refactor assess() | Daniel Friesel | -21/+9 | |
2021-03-02 | ModelAttribute: remove get_fitted(), use .model_function, .model_info instead | Daniel Friesel | -6/+7 | |
2021-03-02 | Move ParallelParamFit to a separate module | Daniel Friesel | -322/+3 | |
2021-03-02 | move ModelAttribute to parameters module | Daniel Friesel | -247/+6 | |
2021-03-01 | Set DFATOOL_NO_PARAM to disable parameter fitting | Daniel Friesel | -0/+2 | |
2021-03-01 | get_fitted: also provide information on static (sub)models | Daniel Friesel | -2/+5 | |
2021-02-26 | refactor param_info, show splits in analyze-archive output | Daniel Friesel | -28/+17 | |
2021-02-26 | add simple decisiontree test | Daniel Friesel | -1/+1 | |
2021-02-25 | kinda proper dtree support (todo: refactoring) | Daniel Friesel | -7/+30 | |
2021-02-25 | pelt_refine: Fix "no sub-states" handling | Daniel Friesel | -1/+1 | |
2021-02-25 | Adjust ParamStats interface in preparation for decision-tree models | Daniel Friesel | -23/+166 | |
2021-02-24 | cluster and re-fit sub-states | Daniel Friesel | -72/+91 | |
2021-02-24 | lut_median_getter: handle param tuples, not just lists | Daniel Friesel | -1/+4 | |
2021-02-24 | do not hard-code substate names | Daniel Friesel | -8/+11 | |
2021-02-24 | treat last sub state as tail / default state | Daniel Friesel | -7/+16 | |
use application-provided total state duration instead of model to determine duration of last sub-state in get_fitted_sub | ||||
2021-02-23 | draft for clustering of parameter-independent sub-states | Daniel Friesel | -91/+96 | |
2021-02-23 | analyze-archive: show substate model quality in table | Daniel Friesel | -1/+15 | |
2021-02-22 | add simple sub-state model accessor and evaluation | Daniel Friesel | -2/+34 | |
2021-02-22 | properly handle static submodel attributes | Daniel Friesel | -27/+36 | |
TODO: fitting and submodel usage in parent | ||||
2021-02-22 | PELT: Increase parallelism | Daniel Friesel | -9/+13 | |
2021-02-19 | start modeling of individual sub-states | Daniel Friesel | -16/+93 | |
2021-02-18 | refactor model generation from Analytic/PTAModel into ModelAttribute class | Daniel Friesel | -309/+210 | |
Iteration over states/transitions and model attributes is no longer hardcoded into most model generation code. This should make support for decision trees and sub-states much easier. | ||||
2021-02-15 | debug output | Daniel Friesel | -2/+4 | |
2021-02-15 | Limit PELT via --with-substates=name_filter=state_or_transition_name | Daniel Friesel | -12/+12 | |
2021-02-12 | PELT: use variable penalty for benchmark repetations | Daniel Friesel | -9/+22 | |
Previously, the first measurement of each by_param entry determined its penalty, and all subsequenc measurements in it were analyzed with this penalty. This lead to some of them coming up with a different number of sub-states and being ignored. Now, all by_param entries are passed to the kneedle / plateau algorithm, and each gets its own penalty. In most cases, this leads to all entries rporting the same number of sub-states, so far less need to be ignored. | ||||
2021-01-11 | evaluate E = P*t model. Seems slightly better in some cases | Daniel Friesel | -0/+16 | |
TODO: Fix send duration model generation for cc1200, should improve eval | ||||
2021-01-08 | model transition power, not energy | Daniel Friesel | -0/+2 | |
in contrast to duration/energy, duration/power are independent | ||||
2020-11-04 | wip | Daniel Friesel | -0/+44 | |
2020-11-03 | analyze number of substates per state | Daniel Friesel | -21/+86 | |
2020-11-02 | Do not pass the entirety of by_nane/by_param to subprocesses | Daniel Friesel | -41/+27 | |
This should speed up analysis quite a bit and also reduce memory usage significantly | ||||
2020-11-02 | Simplify PELT usage. remove kneedle, refactor code | Daniel Friesel | -10/+18 | |
2020-10-09 | add sub-state generation | Daniel Friesel | -1/+6 | |
2020-10-08 | BA Janis import: working detection of optimal number of changepoints | Daniel Friesel | -2/+23 | |
2020-07-15 | Allow state-only PTAModel to be converted to JSON without underlying PTA | Daniel Friesel | -2/+6 | |
2020-07-09 | Remove unimplemented discard-outliers options | Daniel Friesel | -9/+0 | |
2020-07-06 | Remove co-dependent parameter detection code | Daniel Friesel | -25/+0 | |
It doesn't work and is not methodically sound. Decision/Regression Trees seem to be the way to go | ||||
2020-07-06 | Move ParamFit, PTAModel, AnalyticModel to model.py module | Daniel Friesel | -0/+1186 | |