- a193603e
- Event-Object Feature Extraction on OCEL
- 8da05972
- Prefixes and Suffixes filters for Event Logs + Exposition in Simplified Interface
- cbf848ef
- Bug fix BPMN importer
- ff0dfc4b
- Closed security issue within dependencies
- eea18398
- possibility to return the Pydotplus graph inner object in the Heuristics Net visualization.
- 52ddbf75
- support for different attribute keys for the source / target events in the DFG discovery and paths filtering on Pandas dataframes.
- 29bd86a6
- possibility to specify different shifts for the different working days of the week, inside the business hour module.
- f1e124a4
- possibility to move an attribute at the event level in an OCEL to the object type level.
- 0da4c3f6
- custom semantics for Petri net to reachability graph conversion.
- c7c7ed5f185b492f7b6206b04f037a119b80541b
- add "week of the year" option in get_events_distribution method
- 5b5c04874e449bda60463ade6e2cf1a8218e6908
- add prefix/suffix filter for pandas data frames
- 877701fa0e348a5bd58eb84ed984b60292db9f55
- add additional features (useful for instance-spanning constraints) in trace-based feature extraction
- 7359807b60aa3b1ece798d1ef0cdd6a19fac9f6b
- add rebase functionality to pm4py (changing the default activity/case identifier)
- 84742ce331dec418841d99fafb24a82c48c21e7f
- add support for interleaved operator
- d7e232a987e4a0c15e28b9cf2ae6c15ce324031f
- added various additional interaction feature extraction methods for OCEL
- 9caf5597d59ff9eb70879ba42dbfccd9785009af
- add new thirdparty dependency structure in third_party folder
- 74ce9b95
- setting all the arcs of the Petri net visible when there is at least an arc with weight != 1, for coherence reasons.
- 21832737
- inferring the activity frequency from the DFG in a more generic way with regards to the type of the inputs.
- 87fe5afd
- changed tau printing in process tree to string representation, from tau to tau, for coherency with the parse_process_tree operator.
- effce8d8
- changed BPMN namespace in BPMN exporting to ensure compatibility with BPMN modelers.
- 2200a0f5d6d23a1f797199cb834b37e07d8d396e
- add pn to nx converter that returns two dicts for node mappings (pn->nx and nx->pn)
- f9ad1a400846dbdb01f48714df0a3119069a05ea
pm4py.format_dataframe(df)
no longer replaces columns, rather, it copies the data into fresh columns
- 0ad488b1
- Fixed problem in PTAndLogGenerator: silent transitions were added in some context also when the parameter "silent" was provided to 0.
- d07a90873be85d95b15e562aabc6ab1f93b6b109
- removed
pm4py.general_checks_classical_event_log()
- removed
- 034abb0d7a442572f8bd52109ac6ed5cba109d0c
- remove dependency on ciso8601
- c15c8897
- add utility function to convert SNA results to NetworkX
- 8b300dbb
- add several new statistics for OCEL logs
- 8da0f41a
- add frequency-based visualization (using alignments) for process trees
- 54261cbb
- add progress bar to token-based replay
- 225dcad7
- add OCEL schema validators
- da6a4787
- add reduction rules for R/I nets
- fefcd453
- additional support for BPMN functionalities: exceptions and markings
- 417274fd
- add support for feature extraction from OCEL logs
- 5f5ff573
- add filter that checks relative occurrence of a specified attribute
- b82dd92e
- revised implementation of the business hours module, now supports input of work calendars (workalendar package)
- 434e66af
- allow arbitrary arc weights visualized (reported at process-intelligence-solutions#303)
- 76563e4b
- fix bug in process tree alignment that generates NoneTypeError when multiple leaves have the same label
- 3b6800d0
- minor bugfix in process tree playout (reported at: process-intelligence-solutions#305)
- 2eb36ce5
- Bug fix in OCEL importing (timestamp parsing)
- 512c071e
- Resolved security issue in data Petri nets' PNML parsing
- 9b795123
- add converter from data frame to activity/case table
- f28fc490
- add possibility to add the case identifier in the feature table (see: process-intelligence-solutions#292)
- 12b6ec24
- add interleaving DFG visualizer for visualizing inter-process dependencies
- af9c3262
- add first/last occurrence index per activity in the feature table
- 9231a5d7
- add support for conversion of interleaving data structure
- 06f54287
- add support to merge two separate logs using an n:m case-relation table
- 146f49c2
- add the possibility to stream OCEL events and define object-specific listeners
- 573c26c2
- add feature extraction functionality that records the position of activities
- ff62d665
- add case and event sampling to the simplified interface
- d8f71bc3
- add activity-position summary in the simplified interface
- d4011ff1
- add link analysis code for OCEL
- cf72dfe2
- add support for BPMN Gateway direction properties (community contribution by @gregvanhoudt)
- 1c675fba
- fix hardcoded parameter in replay fitness computation (community contribution by @phi1-h)
- 79920a18
- improved string representation of Petri net objects
- 9358fdf4
- minor refactoring for interval detection in event log
- 413cee07
- improvement enhancement for token-based reply (community contribution by @phi1-h)
- 5dccbe61
- fix faulty conversion of process trees to binary equivalent.
- 976cc601
- fix for: process-intelligence-solutions#293
- 1e4f602b
- fix for: process-intelligence-solutions#295
- be629d97
- fix for visualizing multiple tokens in the initial marking in the same place
- a06cc1c8
- fix for the correct use of the triangular distribution on generating process trees
- 51181d6c
- fix support for generating multiple process trees in one go
- 9a0e2be1
- general revision of the process tree generator code
- 32af0c81
- time-stamp based interleaving mining for OCEL logs
- 10dffb58
- support probability visualization in transition system visualizer
- 51c069fb
- add discovery of object-centric directly follows multigraphs
- fa3031aa
- add several filters for OCEL.
- d4747f71
- implementation of OCEL-based process discovery according to Reference paper: van der Aalst, Wil MP, and Alessandro Berti. "Discovering object-centric Petri nets." Fundamenta informaticae 175.1-4 (2020): 1-40.
- 9fbd1c45
- add the support for generic network creation based on a given IN/OUT column in which events are connected if the columns match.
- 2b867f0d
- add projection utility to fetch lists of event attributes
- 43a076c8
- add artificial timestamps to artificial start and end events
- d65f8077
- case attributes are replicated in events of an event stream (for xes input)
- 9075cbfc
- add trace attributes to the interval tree
- 6e26b003
- fixed pandas performance DFG discovery
- 92153184
- fixed parameters usage in simulation packages
- ca6750d4
- fixed hardcoded parameters in dataframe_utils
- 53af01f6
- removed strong dependencies on intervaltree and stringdist
- dcebaf8e
- moving networkx dependency
- f19762ac
- update IMD cut detection to use IM_CLEAN cuts (correct implementation of IM CUTS)
- d5d0b49c
- change the visualization of initial and final markings
- fcc4eeb0
- variant separator (between activities of the same variant) is now a pm4py constant, for increased compatibility with other tools, e.g., celonis.
- 32c396b8
- add sanity checks on event logs objects in simplified interface
- 5b85d5dc
- add utility to parse a collection of traces in string form to an event log
- a87a39c9
- add support for importing XES 2.0
- b43d425b
- add artificial start and end events to event logs
- d22dd490
- add initial support for OCEL 1.0
- 829f091c & 56fca738
- support for business hours in the pandas dfg calculation
- support for business hours in the temporal profiles for pandas dataframes
- support for business hours in pandas df case duration
- support for business hours in filtering for case performance (pandas)
- support for calculating of sojourn time with different aggregation metrics
- 841e3e55
- add etc conformance for dfg models
- 04caa3d3
- add dfg filtering that keeps the dfg connected
- 706d42c0
- bug fix paths filter for Pandas
- c5ecaa4f
- bug fix numeric attribute filter XES (custom case attribute glue)
- 8ba67034
- added random variables that are able to check gamma and log normal distributions
- 1d22d99d
- added dfg -> petri net translation that has unique labels (routing is performed by invisible transitions)
- 004ec93f
- add support for log-level fitness in the alignment output
- 56efe270
- add fitness value for the dfg-based alignments
- d9da1ab8
- add raw performance values for the elements of the performance-based dfg
- 0eeda19d
- when visualizing a dfg without log attached to it, i.e., incoming edges are used to count
- 03ee6b8e
- allow counting of occurrences of activities/open cases/resource activities in a given time range
- ae5a3973
- add various new filtering functionalities
- ac00be2f
- added the specification of Python 3.9.x among the supported versions.
- not suggesting anymore Python 3.6.x
- 816fb4ad
- fixed a bug in the Pandas case size filter (the constraints were not applied correctly).
- 40f142c4
- fixed a bug in the format_dataframe function (columns were duplicated if already existing with the same name).
- 00d1a7de
- reverted stream converter to old variant (in a slightly slower but safer way).
- 991a09d4
- introduce a time limit in the DFG playout.
- ae5d2a07
- return the state of the process tree along with the alignment for the process tree alignments.
- 8b77384f
- refactoring of the calculation of the fitness for Petri net alignments (scattered code).
- d58d34fd
- upgraded Dockerfile to Python 3.9
- 50114175
- resolved issue with the upcoming Python 3.10 release
- 89314905
- security issue in requirements
- 5723df7b
- xes exporter now reports on xes features and xmlns
- 3b632548
- graphviz based visualizations now expose background color as a parameter
- 0592157b
- new dfg playout including performance specification
- 85739ba0
- allow pandas df to be used as an iterable for streaming simulation
- 2fa9993f
- path filter that filters the cases of an event log where there is at least one occurrence of the provided path occurring in a given time range.
- a7ee73a8
- added filter based on rework detection
- c03b6188
- add petri net, reset/inhibitor net and data petri net semantics
- a374bad3
- e88a6546
- 84511628
- fix minor bug in the calculation of the handover and subcontracting metrics.
- 01fd0402
- The
pm4py.view_petri_net()
method now usesNone
as a default initial and final marking.
- The
- 72ed7d0d
- Improved performance of variant discovery of dataframes.
- 9a04357e
- Add rework measurement at the case level in the
pm4py.statistics.rework
package.
- Add rework measurement at the case level in the
- b725ca0b
- add 'between' filter for
pandas dataframes
in thepm4py.algo.filtering.pandas
package. The filter returns subsequences between the two given activities. It creates subtraces for every possible match.
- add 'between' filter for
- 211e3c56
- added local linear embeddings to
log_to_features.util
.
- added local linear embeddings to
- 4b594228
- add support for adding decision points to data petri nets.
- 9261270e
- add support for performance dfg discovery in
pm4py.discover_performance_dfg()
.
- add support for performance dfg discovery in
- 207d69bd
- bug fix in application of the filtering threshold in the IMf algorithm
- d98cbb1c
- changed deepcopy and copy functionality of logs for performance improvement
- f3b78a49
- minor performance optimization in log conversion (log to dataframe)
- 71c0919f
- improved performance for pands -> stream conversion
- f2101a72
- added various additional features in log-based feature extraction
- 41873655
- possiblity to directly get all performance metrics of the DFG elements
- 886b44ea
- detection method for trace-level attributes stored at event level
- d5f9f866
- add transition names to events, based on a given alignment
- 4802e7d8
- add support for importing reset/inhibitor arcs and transition guards
- cc6488f7
- add general support for reset/inhibitor nets
- e805cf5f
- add support for data petri nets
- 1d3a2e7b
- added case termination statistics for pandas data frames
- 50ad39fa
- Fixed blocking issue with properties of Pandas dataframes when format_dataframe is used (case ID column)
- 3708b98f
- Fixed variants filter, when the output of get_variants_as_tuples is used
- Deprecated support to Pandas < 0.25 (due to future dropping)
- Deprecated auto-filters (due to future dropping)
- Different interventions to fix the internal coherency of the project (usage of deprecated functions + missing imports)
- 4964d6ea
- minor refactoring (rename) in attribute statistics querying;
get_attributes()
-->get_event_attributes()
- minor refactoring (rename) in attribute statistics querying;
- 1148f6c0
- use revised implementation of IM and IMf everywhere, deprecate old implementations
- 6750bf3a
- add support for start time and end-time in timstamp conversion
- e24f5b70
- computation of event-level overlap
- 8cec5f9e
- add several case/event level statistic functions at the simplified interface level
- daf74e83
- update imports in feature extraction
- 74be3e3c
- minor bug fix in alpha plus (place that was created was not always added to the resulting Petri net)
- d97b1790
- drop deepcopy in event log sorting (enhances performance)
- 1d4e625b
- revised IMf implementation (more close to ProM / PhD thesis Sander Leemans)
- 20aabd95
- calculation of minimum self distance now adheres to the standard invocation structure
- 598c6ecb
- simplified interface now stores properties (using attr attribute) to dataframes
- 1f7a3fa8
- add computation of rework statistic (cases containing the same activity more than once)
- 32c7d330
- add computation of cycle time (active time of process divided by the number of instances of the process)
- 8187f0e9
- add distribution plots over different time-frames (matplotlib)
- 269d826c
- add batch detection based on Martin, N., Swennen, M., Depaire, B., Jans, M., Caris, A., & Vanhoof, K. (2015, December). Batch Processing: Definition and Event Log Identification. In SIMPDA (pp. 137-140).
- d5326d46
- compute case overlap of a case with all other cases
- 92a70586
- performance optimization for calculation of performance spectrum
- b0fc57c4
- performance optimization for Pandas datetime conversion non-ISO8601 (regular formats)
- c11bab8f
- bug fix in eventually-follows filter
- d3fd1bc1
- bug fix in activity frequency constraints of the log skeleton conformance checking
- d96d9d69
- improved performance of the df-based performance spectrum code
- 499d8a1c
- improved performance of log conversions when (for internal use) deep copy is not required
- 4d679934
- allow the possibility to filter on a trace attribute that has a type date (e.g., does the planned start date of the case fall in a given time window?)
- b7ef36e8
- add properties object to trace attributes (used for internal storage of statistics, will not be exported to disk)
- d7029365
- added some basic ML utilities for event logs, e.g., getting all prefixes of traces, get a train/test split
- 1ec5802e
- new subtrace selection mechanism that gets all events inbetween two given activity labels (uses first match on both 1st and 2nd label)
- 9b65bbd9
- allow specification of business hours in sojourn time computation of the DFG
- 4d529d6e
- generic support for feature extraction
- 908e06d7
- fix error in loop detection of inductive miner
- b7b63e0b
- add internal log conversion in the flexible heuristics miner
- e9d61bdb
- fix minor bug in bpmn model importing
- 52cc0c7a
- fix minor bug in xes exporting (type of concept:name was not checked)
- 9c1a9610
- various old utility functions are now deprecated
- 424c9ad9
- avoid warnings when visualizing long place names in debug visualization mode
- c2a9633e, 52e340b1
- add simple visualization of performance spectrum.
- b6ae4b25
- add simple dotted chart visualization to the simplified interface.
- 6e3a0bac
- add properties attribute to event logs and event streams for storage of custom meta-data that is not exported to xes.
- fb142359
- add version of dfg discovery that adds case-level attributes to nodes and edges
- d902609d
- add basic visualization of events per time and cas distribution graphs
- 766fafa7
- minor refactoring and more generic invocation style for eventually follows-based filtering
- 353c7d6f
- Heuristics miner is now able to filter on edges connecting to/from start/end activity
- d6412339
- Parallel alignment computation can be directly invoked
using
pm4py.conformance_diagnostics_alignments(..., multi_processing=True)
- Parallel alignment computation can be directly invoked
using
- de84e5f4
- add
pm4py.discover_bpmn_inductive(log)
- add
- 9854f62d
- minor bug fix in etree xes exporter avoiding faulty None values
- bfe8fb32
- support non-standard attribute symbols in line-by-line event log exporter
- 3631fe58
- default xes importer is set back to iterparse
- a7ff695a
- large-scale restructuring of the underlying pm4py architecture
- 201879ad
- changed the default maximum number of edges to be visualized in the DFG visualization to 100000 (was: 75)
- 66283964
- sojourn-time-based coloring for dfgs
- 6639d3f3
- organizational mining, e.g.,
pm4py.discover_handover_of_work_network(log)
- organizational mining, e.g.,
- 9c9ca14a
- allow multiprocessing in alignment computation
- 279fd31f
- add prefix tree vizualiation
- 748c768d
- add 'old' pm4py visualization of the process tree as an alternative visualziation
- 408b37a9
- add filter to check multiple ocurrences of certain attribute values in a case.
- 56317d81
- process tree based alignments no longer use trace-based process tree reduction (can still be used through utils)
- c1c1ffc8
- minor optimizations to state-equation based alignment computation
- c95d45c9
- large (internal) refactoring of PM4Py object files and algorithms
- d14d3d27
- added resource profiles to PM4Py taken from Pika, Anastasiia, et al. "Mining resource profiles from event logs." ACM Transactions on Management Information Systems (TMIS) 8.1 (2017): 1-30.
- ab56d899
- organizational mining according to https://arxiv.org/abs/2011.12445; contains several organizational group-based metrics
- 6a77a948
- add serialization and deserialization to various pm4py objects; available through
pm4py.serialize()
andpm4py.deserialize()
- add serialization and deserialization to various pm4py objects; available through
- d1285706
- fixed the consistency (w.r.t ProM) of align-etc conformance results
- c3cde455
- deprecated (moved internally) the evaluation and simulation pacakges.
- a756f1fa
- pm4py.objects.process_tree.pt_operator.py
- 8474507b
- make timestamp and performance-based trace filters inclusive on the boundaries provided
- b6154457
- changed the equals functionality for event logs
- 9eff5646
- classical inductive miner is rebuilt from scratch and thoroughly tested
- efc1c6e8
- changed equals functionality of Petri nets and all their objects
- 02336ff4
- font size is now a parameter of the object (Petri nets / Process Trees /...) visualization code
- 5de03f1e
- added progress bar to all the alignment algorithms
- 24778a7c
- added footprint comparison to simple the interface
- added eventually follows discovery to the simple interface
- added some additional statistics to the simple interface
- b2b1fdc5
- add a faster alignment algorithm for process trees
- b7bc217f
- more extensive support for the OR-operator in process trees
- be04ab2a
- added performance visualization for heuristics nets
- 725f40f2
- added boolean check on whether a trace/variant is fitting w.r.t. a given model (
pm4py.check_is_fitting()
)
- added boolean check on whether a trace/variant is fitting w.r.t. a given model (
- e172977c
- added process tree parsing functionality (
pm4py.parse_process_tree()
)
- added process tree parsing functionality (
- a756f1fa
- the process tree operator class is now embedded within the process tree object definition ( pm4py.objects.process_tree.process_tree.py)
- 1a5c080c
- fix for timestamp conversion of dataframe formatting
- 19c615e1
- fix bug in process tree exporter
- change visualization of process trees (similar to PorM)
- 0e61f4b2
- evaluation.soundness.wofland and evaluation.wf_net
- 0e61f4b2
- woflan and wf-net checks are moved to algo.analysis package
- 2e158ec4
- minor improvements for A* performance
- d550f777
- various renamings in the simplified interface of pm4py, several methods are deprecated.
- 65ef822c
- generic support for the marking equation
- generic support for the extended marking equation
- 92ba4aa7
- variants can now be represented as a tuple of activities, rather than a single string
- ee11545a
- fixed importing names of invisible transitions as stored in
.pnml
files
- fixed importing names of invisible transitions as stored in
- 5efff284
- handle warning messages thrown in the heuristics net visualization
- 91b494ad
- simple process tree reduction that removes parts that are guaranteed not to be needed for the alignment/replay of a trace
- f75ecff3
- thread-safe implementation of
dict
for streaming based process mining
- thread-safe implementation of
- 03d176f9
- implementation of the Heuristics++ Miner
- 32443759
- add support for using
redis dict
for streaming
- add support for using
- ee545f40
- add additional check to timeout for the memory efficient implementation of A* approach for alignments
- a2a3f281
- fix usage of integer values in pulp solver rather than binary variables.
- 6ba4322f
- fixed conversion behavior lifecycle to interval logs
- dd3e22da
- pass the flow name to edge attribute of graph in BPMN. Reported by @YJDave; see: process-intelligence-solutions#202
- ce240905
- fix for GLPK ILP when no solution exists. see: process-intelligence-solutions#204
- 54e38ac8
pm4py.soundness_woflan()
is now deprecated
- c847e39c
- bpmn graphs are now multi-di-graphs. also see: process-intelligence-solutions#203
- 54e38ac8
pm4py.check_soundness()
replacespm4py.soundness_woflan()
- aa91fdf7
- add typing information to
pm4py.conformance.py
(containingpm4py.conformance_alignments()
etc.)
- add typing information to
- 5d7890b2
- added
insert_ev_in_tr_index()
utility to dataframe utils: possibility to insert the index of an event inside its trace (e.g. the start event gets 0, the event following gets 1). Allows us to quickly filter on prefixes directly at the dataframe level. - added
automatic_feature_extraction_df()
utility: possibility to extract the features of an event log directly starting from a dataframe. Also, an utility for the manual specification of the columns that should be considered in the event extraction is provided.
- added
- f64c9a6b
- add option to infer concurrency between to activities in a 'strict' manner in the log statistics.
using
srict=True
, implies that an overlap of '0' zero is not considered concurrent. also see: process-intelligence-solutions#201
- add option to infer concurrency between to activities in a 'strict' manner in the log statistics.
using
- c0083f68
- implementation based on Stertz, Florian, Jürgen Mangler, and Stefanie Rinderle-Ma. "Temporal Conformance Checking
at Runtime based on Time-infused Process Models." arXiv preprint arXiv:2008.07262 (2020):
- add temporal profile discovery
- add offline conformance checking based on temporal profiles
- add online conformance checking based on termporal profiles
- implementation based on Stertz, Florian, Jürgen Mangler, and Stefanie Rinderle-Ma. "Temporal Conformance Checking
at Runtime based on Time-infused Process Models." arXiv preprint arXiv:2008.07262 (2020):
- 4d3cf81c
- support serialization of all pm4py visualizations
- 453805b4
- compute alignments using edit distance (requires two sets of traces as an input, one represents the log, one represents (a subset of) the model behavior)
- 1231f518
- strip text read from nodes in bpmn importing
- 0bc1b330
- add type checking for bpmn conversion; i.e., if the input is already bpmn, it is returned.
- ea0c7e54
- fix consistency in obtaining the case arrival statistics in
pmpy.statistics.traces.log.case_arrival
; was median, changed to mean. also see: process-intelligence-solutions#200
- fix consistency in obtaining the case arrival statistics in
- 51be0910
- set
stream_postprocessing
default value back toFalse
fordataframe
tostream
conversion. Columns containingNone
values are no longer filtered by default ( compliant withpm4py<=2.1.2
). also see: process-intelligence-solutions#199
- set
- 8976ad45
- drop the explicit dependency on
numpy 1.19.3
- add explicit dependency
pulp<=2.1
- drop the explicit dependency on
- 1231f518
- support
sequenceflow
operator node in bpmn file import
- support
- 62618eeb
- explicitly blacklist version 1.9.4 of
numpy
in the requirements.txt file.
- explicitly blacklist version 1.9.4 of
- 35f2278a; 89c5f13b; 6a3579bc; 65fc182b; fa4448a6; c4e44311 c456c681; 6c6d96cc; e3770281; f091c43e; 6a20cf17; 69eb1ae7;
ca780326; 36cb3963 e4f3b16f; c9f80d1f; 94c5a6e0; a713ef3d:
- add fall-back to time-range filter if incorrect argument is passed
- fix the copying of the 'meta attributes' of the filtered event log for the start activities filter
- fix the copying of the 'meta attributes' of the filtered event log for the end activities filter
- fix the copying of the 'meta attributes' of the filtered event log for the attributes filter
- fix the copying of the 'meta attributes' of the filtered event log for the variants filter
- fix the copying of the 'meta attributes' of the filtered event log for the directly follows filter
- fix the copying of the 'meta attributes' for event logs in the ltl checker
- fix the copying of the 'meta attributes' for event logs in the timestamp filter
- ffe29353:
- create event log object before starting the parsing of XML file (in ITERPARSE_MEM_COMPRESSED)
- 8f2d899a
- allow to specify a cap on the number of times we visit the same marking in the extensive playout
- allow to return the firing sequences of extensive playout instead of an event log
- b707377b
- allow to return the firing sequences of basic/stochastic playout instead of an event log
- 9782f522
- extended exception management in streaming algorithm interface: release locks if exception happen
- 0a741566
- support importing of bpmn files that do not describe a name for gateways
- 583825d8
- refactored variant-based filtering: added top-K and coverage percentage
- ba073f54
- extended DFG filtering
- 8ebda3b1
- exploit variants in the extensive footprints conformance checking
- dc754c78
- change range(s) of timestamp filters to be inclusive (<=) rather than exclusive (<)
- b5cb7f0d; f1c0f037; 960d40e9
- fix naming consistency in the filtering interface of pm4py.
- f9f894ed
- create an iterable that permits, theoretically, to iterate over the infinity of traces admitted by a DFG
- the main
apply()
calls the iterable and stops with the usual criterias - the
apply()
can also return the variants of the log
- 1c3666b7
- minor refactoring of the filtering simplified pm4py interface
- 8b5dee65
- add lambda-based filtering and sorting to simplified pm4py interface
- 388348f2
- bugfix
remove_flow
BPMN function
- bugfix
- 3bd4fe0a
- bug fix - DFG visualization needs deepcopy, otherwise it can remove element from the underlying DFG
- 92fde8cf
- bug fix in Petri net playout in stop criterion
- 882468e1
- compatibility with pulp version 1.6.x
- 41ed5720
- deepcopy of inputs: since the dictionaries/sets are modified, a "deepcopy" is the best option to ensure data integrity.
keep_all_activities
parameter in paths filter: decides if all the activities (also the ones connected by the low occurrences edges) should be kept, or only the ones appearing in the edges with more occurrences (default).
- 63ccc055
- IM and IMf: removed dependency on
pm4py.algo.filtering
package
- IM and IMf: removed dependency on
- 8a5788fa
- more advanced to bpmn conversion in the simplified interface
- 66e0c074
- refactoring conversion parameters log->stream and improved stream compression
- 73054b04
- improved performance of the line-by-line xes importer
- increased XES-standard conformity
- e6136ce5
- improved performance of the xes exporters
- increased XES-standard conformity
- progress bar for exporting enabled
- 3b692b33
- added parameter to enable/disable progress bar in importing in ITERPARSE
- compression from file - XES compression moved from general entrypoint to single variant
- 68ff7d00
- more efficient importing of .xes.gz files
- 4aad427e; 67a42d23
- get predecessors and successors of a dfg node (in dfg utils)
- 0888ab26
- added minimum trace length in process tree extensive playout
- 6a946fbd
- allow to compute alignments directly on the dfg using dijkstra
- 50722bb8
- DFG playout including Markovian probability of traces
- 2fb0b807
- support nodes with the same label in BPMN layout algorithm
- 96a7681c
- significant memory footprint reduction for iterparse-based event log importing
- 8270a46e
- significantly faster 'line-by-line' xes exporter
- aae4be33
- advanced DFG filtering in (activities percentage, paths percentage) ensuring reachability from start and end
- d38f4c97
- number of occurrences as a start or end activity is visualized in the DFG visualization
- 97cc315c
- add (de)serialization functionality for PM4Py objects
- 43e7ce9e
- full conversion of WF-nets to BPMN models
- 029d30f5
- add visualizing and exporting BPMN models
- f76dd379
- added conversion to simplified interface
- f18c3e17
- downgrading
numpy
to 1.19.3 for Windows 10 2004 compatibility problems - skip blocking
scikit-learn
installation for Python 3.9
- downgrading
- f6cdf1a9
- hotfix problem in XES importer reading of parameters
- a0a7fd09
- bug fix in the inductive miner: sequence cuts were to maximal, leading to underfitting models (involving too many skips)
- 5b32725a
- fix use of deepcopy in event log conversion
- 33103b5c
- fix get-variants behavior for event log / df (yielded different results)
- 0138f93d
- deprecated
pm4py.write_csv()
; conversion and subsequent pandas export should be used.
- deprecated
- 3b8fcd4a
- apply invisible transition reduction on Petri nets obtained by Heursistic Miner
- 6afd8ab9
- support for importing/exporting BPMN files;
- supported elements: tasks, xor, and, or gateways
- conversion of BPMN to WF-net
- conversion of Process Tree to BPMN
- 848a1610
- conversion of dataframes to event streams / event logs now detects the use of XES Extensions
- 8874f7b4
- add statistics on concurrent activities
- discover eventually follows relations with concurrent activities
- 4c115dbb
- fix bug in hash function of traces
- 0dd8f28e
- fix bug in the alpha+ algorithm (was adding artificial start/end to the underlying event log)
- commit c51b1f02
- parametrizing size of the thread pool in streaming package
- 8e07d847
- add (unbounded) dfg discovery on an event stream
- 16e3f7f5
- integration of the wf-net to process tree algorithm described in https://doi.org/10.3390/a13110279
- 1a26f678
- allow creation of 'live' streaming objects directly from xes and csv files
- b08564ed
- start minor support for interval-based process models
- 36cb7130
- added support for generic dictionaries in streaming conformance checking
- 81579e19e
- relaxing the importing of some dependencies (pandas, pulp, graphviz, intervaltree) to make a basic set of functionalities of PM4Py work even without those dependencies
- ed45eafc
- fixing circular dependencies issues and added partial compatibility with Python 3.4
- 1a8f9281
- bug fix in
pandas
case size filter
- bug fix in
- 7a89d4dd
- remove deprecated factories
- 62801513
- improved performance of token-based replay with duplicate labels
- f408a181
- add streaming token-based replay and footprint-based comparison
- fde08b03
- output conformance checking diagnostics in a dataframe
- a9c5aa34
- compatibility with Python 3.9 (limited to windows)
- 2d4a8b67e25164838030dca709816918ddbf9279
- fixed dependencies on factories. fixed some release notes. fixed some deprecations.
- 0d75241676a736919952e5c9ab96ab2be4c71046
- added utility for trace attributes filtering
- d8433ac1bbb57082b19b3b97bc726b353096c76e
- better specification of 'stable' requirements
a6fde4ccfdf0465d8919d3c72d4400bce8d4ab0e
- fixed some deprecation warnings for external libraries
- a5befe62f84af0f3dbe734c713cc8c2c1fc22a04
- added WOFLAN to the simplified interface
- 7bf3aea0b4638f6eceaa192ec37e17370e47f560
- fixing a parameter issue on stochastic playout
- 13cd96da7fe22427839f1b21dd38a5ac8ff8c231
- provide alternative visualizations through Matplotlib of the main visualizations
- e7e3dfa11264a7e737ca7b65165eb7f719817f59
- add utility to remove an arc
- c42bad174247eb13134e0cb4acb0703c301aa869
- fixed compatibility with Python 3.5
- 9e926a5d1a373d124a041c72f909eb34083f4649
- removed the old soundness check
- 526342e3e96c7386dc7451ac0d9ec74c76b86b4f
- performance improvement of the 'less memory' alignment variant
- 83c71eaaa9bb4ee55e2f773dd3fc9704c6324704
- generalization of the heuristics miner visualization
- 7e7d9ba9e9b4f94d9801f5c8df3c5e4b3afd5971
- improved speed of event log sampling
- 11d06ba2a44fd7d0a83e439b6615befad5e3c4ba
- faster process tree playout (recursive, ignoring PT statespace)
- ce3a8609013bd7f5890b52cb44e155b57b2fc1c0
- allow adding (arbitrary) data to transition system
- 30b9059fac3d59abac1c973637bd66388d7dd6db
- bug fix in parameter parsing of log conversions
- 3c9ed4b5e1860e2494d3b68668e3e1021ee59585
- apply invisible transition reduction in process tree to Petri net transformation
- 26ac905fed4703724805f8fb1b5daa21475df147
- minor optimizations in process tree alignment approximation algorithm
- c5b5ffdd51cd832be14d3078948e29f88f6c2673
- add time-out parameter to decomposed alignment calculation
- 4a5fb9ff5ab395529062ce92747b9624a1799b0e
- add bipartite graph matching for correlation mining to obtain exact matching results
- acc46269040a3b60bdcc6f5afe9ea55ddb5549e1
- minor performance improvement of process tree playout
- 5dd0fe17d5a07a2096afc647306a5673d86508ba
- apply trace-level directly on pandas data frames
- 4c35335da6a6d4419b68a49af3cbd358eca8881d
- add 'COUNT' functionality for log/dfg statistics, e.g., in how many cases does act 'A' appear?
- 3bb5125f1849985ec9e2ba7962a287a25ede8e43
- new implementation of the (classical) Inductive Miner, based on the PhD thesis of Sander Leemans. The
implementation covers the following fall through functions:
- empty trace
- strict tau loop
- tau loop
- activity once per trace
- activity concurrent
- flower loop
- new implementation of the (classical) Inductive Miner, based on the PhD thesis of Sander Leemans. The
implementation covers the following fall through functions:
- 1639d7a17660b4641a58cb435f7061d4168ce422
- implemention of 'WOFLAN', based on the PhD thesis of Eric (H.M.W.) Verbeek.
- 76df914ad48b93a72127b0d79e5547669d3148db
- fixed bug in less-memory alignments when im == fm
- fc724c4c9e00a388b94efea50be46412c1351bc5
- refactoring correlation miner
- increased scalability of the correlation miner
- e18021f06e4431628a84b441acf136d44c9baed6
- revision/enrichment of simplified API
- 9cfd4183968118f488c95ab0bbd7e3e9b66ba355
- alignments approximation on process trees (https://arxiv.org/pdf/2009.14094)
- f9a898da47367f86bf616a7bc8dc796be1b3e440
- discovery of log skeleton from the list of variants
- c5fa9dee9bf7e4eed56785eadc423b59214d3df0
- compatibility with latest IPython API
- 616623b358d1e1a9aa789f43e2c8cf7670ee4d79
- clean-up in linear solver functionality
- bug fix in usage of PulP result vector
- 4616952ee5c184d35d26e986b298e1e586b38b3f
- minor refactoring of the alignment code
- 3477e5345068018fa6aa91e04b6da4184e4d3c94
- update of verions of package dependencies
- cba7a2e574aa15709d0a2dda35f8a27da9200f42
- [beta release] simplified 'pythonic' method invocation (see
examples/simplified_interface.py
)
- [beta release] simplified 'pythonic' method invocation (see
- ade44e6d78551c454b477f36cbe3248a5d4e6c8b
- added the 'correlation miner' to PM4Py (https://is.tm.tue.nl/staff/rdijkman/papers/Pourmirza2017.pdf)
- 9564b54a7111916c7bc3d6f1e4c56f3829978efe
- add visualization of conformance checking results on process tree
- 988258a2c655d17201c5b1a7120d19ab5954dbe3
- added 'extensive' playout for process trees (for footprint comparison)
- a31467a560b940a2cb426751e032b4ea79bbb29f
- add cast to Event class in event conversion
- 0febe4eda5772685498da2d79d144effdf61ce20
- fix in alpha miner; conflict check yielded self-loops in some cases
- 844459e366e77ebf88a8d01728064048d71a3e6f
- remove
pyemd
from requirements as it requires a full C studio installation; can be installed and used * optionally*
- remove
- 6cdf4483548027420b0a54c681a3ecfd515ab8f3
- bug fix for de/recomposition approach to handle duplicated labels
- 58149e7f085bbedb47e31d58d10e37d686e46a72
- small fix in importing of parquet files
- b08675fe8c03207c4701d2eb8b91c234e3ddbe1b
- fix in IMdf; the noise threshold was overridden before
- 772ea5d5af9ab2fa3a424b91670d81487ab2667b
- minor fix for timestamps that end with symbol 'Z' (Zero timestamp)
- 1e13cbdb011dff175850c4c665283bf60e0db84c
- fix Scipy dependency for tree generator
- 921328cd06d8870ecaa601d53cfde30a6015aec7
- parameterize simplicity metric to use a given 'K' for reference
- 8403093c3e18c1493f4fd72bc83914dadb03e4c2
- change the color scheme for net comparison
- 143f60a0b4fb265ba5e9d997efddf8f8d66458c3
- add support for process tree importing (.ptml files)
- 4b9132f2835f574d9e7e5af113e5fc96bd821ccf
- add new A* version that encodes the states of the state-space differently (up to 10% memory reduction)
- 93e6e84c4ea78afc562f4ba6cee42de8718eb4cb
- add 'stochastic' playout, i.e., use token-based replay on a given log to guide the playout of a given Petri net
- 107797c5d1be1d742e6522a14231f4f9fce1cb38
- add EMD-based process model evaluation ('Earth Mover's Distance' between log/model)
- 1e02596bc90b6fc22a8c89e80567bdddc2c8f1b7
- add process tree based footprint computation
- f05ba60e2faa28b2f2e2d7ef65e0ac415ae298e8
- improved memory performance of Dijkstra-based alignments, restriction of model-move scheduling is applied (only one of many is chosen)
- eaed442d10a5b4c9eead3966cef473e2c3f0d61b
- improved efficiency of reachability graph computation
- a42d0f087c94cb9e7914532f7ae0228af3a7f9ce
- add 'extensive' playout, which allows to generate all traces in the language of a model up-to a given length.
- eaed442d10a5b4c9eead3966cef473e2c3f0d61b
- added footprint-based conformance checking
- b99da02411259897f1df52f9a9a62153228dab8a
PuLP<=2.1
is forced (for now, due to excessive console output of v2.2.)- new version of
Pyvis
requires additional dependencies (Apache, BSD, MIT licenses), which are now specified
- 1278713e2209d7b1e2287afbd9d80097db028617
- minor fixes in (to-be-deprecated) Parquet importer and exporter
- 59630ed8456d9e7842e05fee4c53ff1a3d1389a0
- improved memory performance of Dijkstra-based alignments
- 98e6d6d6219cceeeb2421392e9981ec808d2a468 (Merge Request #89)
- Support importing and exporting of '.dfg' files (similar to ProM 6.10)
- 7ca768b7fcd1eea468bcdb1c1e8c9397676b511a (Merge Request #90)
- add a detailed list of all third-party dependencies (README.THIRD_PARTY.md)
- e0685638c6c0c171583108e23f752ebef70d033e (Merge Request 94)
- new version of Dijkstra based alignment computation
- d51430106dea70473777c6868a3676c027faf607
- fix paths filter for Pandas dataframe when dataframe is not sorted
- c7a76e0aaffb5dbcad769b2fbc52fd8c03703769
- cleanup of the petri net impoprter/exporter, i.e., stochastic/layout information is now stored in the 'properties' object of places/transitions/nets
- 00d7c405628033e8f383a845e61a84205f7bbc83
- color scheme of the log-log comparison (pm4py.algo.enhancement.comparison.petrinet)
- e84c7dc88907b408b4caf97524dd69a378e1859e
- update the README.md file
- 332de04112cf780cbc711553e34ca4b835ee8425
- add parameters object to transient analysis call to dfg learner
- f0976015c2b5b7384135855e75b36f69fb44a4db
- add final marking to the basic playout of Petri nets
- e0d43c6dd5ee2492eabfd14c507b430088ec9cf0
- allow prefixes in petri net attribute names (petri net importing)
- 095f35a3bf9f9c3bea0d518c72c76380f38b77b2
- add support for (correctly) importing lists from .xes files
- 112dc3fc56f0d5e3f3ae0c737a5b0b001be31ba5
- typo in Petri net weight (reporter: dominiquesommers)
- 4a74ef159d03d3cf3ca43bd7c99f5f97da16baf8
- problems in the evaluation of replay fitness of recomposed alignments
- 1f40a17dcade522d308b56a9f53b40c5b8a95dfc
- cleaning circular dependencies. relaxing some absolutely strict dependencies for some border-line configurations.
- 9780ef609db7bbb24323d3584abc2338f24252d3
- copying of Event/Trace/EventStream/EventLog objects (reporter: M.Pegoraro)
- 51802f0828a6d207a5185c06a71bdb3a9faa1a46
- problem with sampling log (reporter: jacksbrajin)
- b1f16bf087691181adcf5616aa85e3e454a7169c
- fixed problem with the discovery of 2-loops in the Heuristics Miner (reporter: czwilling)
- c96dcb16064ed156a1f48075445113bb5afc6264
- fixed
get_variants_acyclic()
function in petri utils (authorship: M.Pegoraro)
- fixed
- b4cea4be6fe6ff90f58c405f9f0acb4dbac973f4
- factories are deprecated
- b4cea4be6fe6ff90f58c405f9f0acb4dbac973f4
- factories have been renamed
- e8f250fec6781c088da8204923ad8817f1b5d689
- decision point mining with alignments (for any scikit classification approach) is now integrated.
- 4ff811ab80b5df15e5ec1da082b4f76c72dcf684
- trace attribute hierarchical clustering (MSc Thesis Yukun Cao; FIT/RWTH)
- 4fca12c1608818348bbd51e390140a8f7e79d7f6
- alignments decomposition/recomposition
- f8d52aa69a3d720d45a93b35c79b4241bb8f7691
- possibility to provide the final marking to the playout factory
- 9a5a64ea2941d69be07ee0e93b771c80f5820166
- hash functions for event log objects
- 9f76d2e61dbc1cc74fccafac6f60d9ce69b7c791
- added
table_to_stream()
auxiliary function in dataframe_utils
- added
- 531b767d85bb4c95996ae0c9644a958f75aad120
- utilities of Petri nets (acyclic net variants (M. Pegoraro); strongly connected components by NX graph)
- c9cebbfab9c82bf8edbe8851a000fd1b1f31f8be
properties
object to all members of the Petri net class (including transitions/places/arcs)
- ef3b4b62fd186df46236a8af9aa890358dbcd1bc
- problem in the generation of logs from process trees
- 548c57a6d2340dcaba7ef11464ebe193f8fb9c5c
- filter by variants percentage
- c49a9c441feb65a74d5c5da774fdda79295665cc
- problems of the token-based replay with the count of tokens at global and local level
- f554aec318717ee1fb7f81c4e0acbd6da7e7bc34
- continuous time Markov Chain steady-state analysis
- e05145de972944f211f2763656eb6e41aa64e0b5
- revised process tree fold and tau reduce functions
- c3f66b8cb8667c2f204a1da899a216656386c2fa
- ignore comments in .pnml files
- 31e1cd29437d6b183357bbc5c103131484d390b3
- problematic dependencies ortools and pyarrow in the project (when installed, can be used, but no more required by default)
- f554aec318717ee1fb7f81c4e0acbd6da7e7bc34
- converting a performance DFG to a Q-matrix for transient analysis
- cc766164f2397fe2cb33f8278372469e25cfecd6
- business hours module supports full-days shifts (e.g. from 0 to 24)
- cc766164f2397fe2cb33f8278372469e25cfecd6
- backwards state space exploration (supports duplicate transitions) for token-based replay
- 1b3c32916fe0ccbc4d3f73a44cc68e2ca83a810e
- Visual log comparison on a Petri net by plasma coloring
- 6f320562a836cd949dbbe6ec7751f8f9514b01ec
- parameter object for calling alignment code directly
- 471f414820c7b3f919aa97ccdc684d40cf132b3c
- dependency problem with ORTools and PyArrow
- 157fadffd3b953c105e67f4909547f097d77ea0a
- reduce inernal (cyclic) dependencies within PM4Py
- f38b6089b7eb6a4518a9c33e9775120874352657; af1328e2dad82f0a059e00942167a29cb918c85f;
e8e1ab443f2dedb2cd348f35bc49eed412d66e1d
- hash and equals function of process trees
- 229bd7ed78ea80aefbb6c7fcfa173edda682c1c7
- refactoring of the Monte Carlo simulation.
- extend the simulation to support arc weights that are provided by the replay (e.g. informed transition pickup)
- extend the simulation to support more than one resource per place through semaphores
- maximum execution time per simulation thread
- general refactoring of the code
- introduce logging information about the simulation
- improving documentation
- improving clarity of the code
- refactoring of the Monte Carlo simulation.
- f5132302e06aa49f26ab3264bb3147d0660a11c0
- performance of log generation from process trees
- 9e64f2635123aa5c3146fb2ad03863cbf93175df
- fixed incongruency in variants statistics (was there for log, not for dataframes)
- increased coverage of tests, and made some of them more lean
- introduced some additional tests on the new functionalities
- updated Dockerfile
- updated setup with new packages
- removed remainings of the SIMPLE algorithm in the tests folder
- 382162c648fb8e32b0145a6ff9427af8a7fb39fa
- fixed parameters initialization to make it uniform for the rest of the project for all the versions of IMDF
- added missing documentation in factory (was there only in the versions)
- moved log conversions from factory to specific versions (since they are indeed version specific, if later we include the log version)
- removed some useless calculations on the new DFG based versions (were there, but never used!)
- introduced two new methods, apply_variants and apply_tree_variants, that are able to apply inductive miner from a list of variants
- separated DEFAULT_VARIANT from DEFAULT_VARIANT_DFG (indeed, the first in future versions may become the log)
- increased number of tests in the tests/ folder for inductive miner
- 502c5d722d483c567f0e08c16da0bb4c87a94e36
- remove the empty performance spectrum visualizer folder
- e8b030afd37559c93f79e1dd030fdc540aa62135
- provide fast-parquet library as an alternative to read/write parquet files.
- 557472bc78900d90beb0757279ef29b89aa410b1
- integrate LogSkeleton for process discovery and conformance checking
- 90ba7bf1495fdac7ccef3112efa95687c46a5dd1
- importing and exporting of Petri net weights
- f571ec65ca544a9322b89cf96299d03da65de5a0
- import DFG to PN
- process tree parsing
- dca5cc602ba8381e5d4e265341dfc5a5292a80c0
- problem with alignments when transitions have empty preset
- 68643cb109503e54787f98eb8f40650e37aa151e
- process tree hashing
- 9d27f132d40638933be00f9d178e9a1167d36166
- conversion between the log types
- 945fd64e481c0fbf020da7f71cb1a7974ae9629c
- make visualization deterministic for process trees, Petri nets and DFGs
- 15be58abb314c679ddf3b65fa6832c680768c413
- generic parsing of dates (removed strict dependency on ciso8601)
- be0a282be033765c9d1d7f1a7ba541a11c046834
- matplotlib backend settings
- ed42182f32eba37df71d9f466ad165036f8d1086
- full support for numpy v1.18
- d1a418f8fa2513a469149383fe69df9a4e6fea06
- full support pandas v1.0
- 6908a34c73c74c42aac3ddf31b964fcae680919e
- inductive miner sequence cut detection
- 881f1fa4e76b9e35c2d80ab7d241183b2d6871c0
- 'Best effort fix' for graphviz based visualizations
- programming error in the alignments code
should_close
for XOR node in PT generationexecute_enabled
for parallel and OR node in PT generation- process tree children setter function
copy.copy
for parent nodes of childs in log generation from process tree- windows platform reuquirement for ortools/pyarrow
- recursion depth in token-based replay invisibles exploration
- consider only fitting traces in the Align-ETConformance count
- consider all the optimal alignments (as described in the paper) and not only the first one
- moving of utilities and making the search function using Dijkstra (since we are looking for fit paths) instead of A*
- LTL checker: making existing filters more performant (A eventually B eventually C)
- LTL checker: introducing A eventually B eventually C eventually D
- LTL checker: introducing time boundaries for each arc of the eventually follows relation
- bug fix in the 'escaping edges' based precision
- Pyarrow serialization integration (supporting in-memory and to .pkl file serialization)