summaryrefslogtreecommitdiff
AgeCommit message (Collapse)AuthorLines
2018-03-08do not hardcode analyzed model attributesDaniel Friesel-38/+59
2018-03-01add some data documentationDaniel Friesel-0/+100
2018-03-01re-add argument support, starting with --ignored-trace-indexesDaniel Friesel-21/+46
2018-02-23regression_measures: add R^2 coefficient of determination score functionDaniel Friesel-1/+8
2018-02-22skip functions which failed to fitDaniel Friesel-6/+7
2018-02-22proper support for arg-dependent model functionsDaniel Friesel-18/+58
2018-02-21decrease _try_fits IPC overheadDaniel Friesel-1/+3
2018-02-21failed attempt at more efficient parameter statistics computationDaniel Friesel-44/+60
2018-02-20run fit checks in parallelDaniel Friesel-87/+157
2018-02-19Include function arguments (if any) in LUT keysDaniel Friesel-48/+70
2018-02-14move quality result printout to analyze-archive.pyDaniel Friesel-31/+48
2018-02-14proper paramfuncton support (no crossvalidation yet for assessment)Daniel Friesel-1/+18
2018-02-12determine functions for parameter-dependent model attributesDaniel Friesel-27/+204
2018-02-06determine most likely analytic representation of a parameter influenceDaniel Friesel-0/+157
2018-02-06simplify model generationDaniel Friesel-22/+2
2018-02-06calculate parameter dependenceDaniel Friesel-9/+105
2018-01-29weed out clipping measurements as early as possibleDaniel Friesel-5/+13
2018-01-29add parameter look-up table modelsDaniel Friesel-7/+59
2018-01-29add some testsDaniel Friesel-5/+194
2018-01-26support analyzing multiple measurements at onceDaniel Friesel-39/+48
2018-01-26Only show SMAPE if it could be calculatedDaniel Friesel-6/+16
2018-01-26use median for static models, not mean. d'oh.Daniel Friesel-2/+20
2018-01-25print model data, assess model qualityDaniel Friesel-23/+80
2018-01-25Re-implement basic state/transition-based analysisDaniel Friesel-20/+105
2018-01-22Do not rely on self.setup (prepare for multi-measurement support)Daniel Friesel-9/+10
2018-01-22Implement preprocessing without perl script hackeryDaniel Friesel-15/+186
2017-09-18Remove support for XML modelsDaniel Friesel-168/+4
2017-09-18Support specification of transitions in AspectC++ annotationsDaniel Friesel-44/+123
2017-09-06Add support for custom baud ratesDaniel Friesel-1/+3
2017-09-06Support new UART "endl" implementation in KratosDaniel Friesel-1/+1
2017-06-01UNINITIALIZED always has ID 0Daniel Friesel-0/+4
2017-06-01Model: initialized transition (relative) energy to 0Daniel Friesel-7/+13
2017-05-30calculate Spearman rank-order correlation coefficientDaniel Friesel-2/+31
2017-05-30Model: DriverEvalThread_{class}Daniel Friesel-2/+2
2017-05-29DFADriver: DriverEvalThread_{class} für Benchmarkwerte per KlassensliceDaniel Friesel-7/+9
2017-05-22support benchmark values via attributes (for int) and slice variables ↵Daniel Friesel-5/+33
(complex data) Can be made less of a hack as soon as AC++ supports attributes with arguments
2017-05-19Model: Support custom extra transitions by state (e.g. VCOM toggle in ENABLED)Daniel Friesel-0/+8
2017-05-18don't get confused by polymorphismDaniel Friesel-1/+26
2017-05-18Fix JSON generation from class/attributesDaniel Friesel-3/+12
2017-05-17Use JSON models instead of XMLDaniel Friesel-296/+141
JSON is easier to (de)serialize and the manual editing possibilities of XML don't matter after the switch to model file autogeneration using attributes and aspects
2017-05-16autogenerate simple models (without args or parameters) from AC++ attributesDaniel Friesel-7/+137
2017-05-15perltidyDaniel Friesel-239/+270
2017-04-28support name suffixesDaniel Friesel-1/+11
2017-04-25get_state_power_with_params: support param = 0Daniel Friesel-1/+1
2017-04-25DFADriver: show warning in high-power state with probable energy counter ↵Daniel Friesel-0/+12
overflow
2017-04-21allow setting static params via LUTDaniel Friesel-4/+64
2017-04-21preparations for trace evaluationDaniel Friesel-92/+99
2017-04-20remove 'dfatool validate' supportDaniel Friesel-158/+2
2017-04-20support both logTransition (with trace logging) and passTransition ↵Daniel Friesel-4/+4
(accounting only)
2017-04-19fix crossvalidation outputDaniel Friesel-10/+10