def calculate(heu_net: HeuristicsNet, dependency_thresh: float,
and_measure_thresh: float,
heu_net_decoration: str) -> HeuristicsNet:
"""
Calculates the dependency matrix and the AND measures using the Heuristics Miner ++ formulas
Parameters
----------------
heu_net
Heuristics net
dependency_thresh
Dependency threshold
and_measure_thresh
AND measure threshold
heu_net_decoration
Decoration to use (frequency/performance)
Returns
----------------
heu_net
Heuristics net
"""
heu_net.performance_matrix = {}
heu_net.dependency_matrix = {}
heu_net.dfg_matrix = {}
for el in heu_net.dfg:
act1 = el[0]
act2 = el[1]
if act1 not in heu_net.dfg_matrix:
heu_net.dfg_matrix[act1] = {}
heu_net.dependency_matrix[act1] = {}
heu_net.performance_matrix[act1] = {}
heu_net.dfg_matrix[act1][act2] = heu_net.dfg[el]
heu_net.dependency_matrix[act1][act2] = -1
heu_net.performance_matrix[act1][act2] = heu_net.performance_dfg[
el] if heu_net.performance_dfg else 0.0
for act1 in heu_net.activities:
heu_net.nodes[act1] = Node(heu_net,
act1,
heu_net.activities_occurrences[act1],
node_type=heu_net.node_type)
# calculates the dependencies between the activities
heu_net = calculate_dependency(heu_net, dependency_thresh,
heu_net_decoration)
# calculates the AND measure for outgoing edges (e.g. which activities happen in parallel after a given activity)
heu_net = calculate_and_out_measure(heu_net, and_measure_thresh)
# calculates the AND measure for ingoing edges (e.g. which activities happen in parallel before a given activity)
heu_net = calculate_and_in_measure(heu_net, and_measure_thresh)
return heu_net
def calculate_dependency(heu_net: HeuristicsNet, dependency_thresh: float, heu_net_decoration: str) -> HeuristicsNet:
"""
Calculates the dependency matrix using the Heuristics Miner ++ formula
Parameters
--------------
heu_net
Heuristics net
dependency_thresh
Dependency threshold
heu_net_decoration
Decoration to include (frequency/performance)
Returns
---------------
heu_net
Heuristics net (enriched)
"""
for act1 in heu_net.activities:
if act1 in heu_net.dfg_matrix:
for act2 in heu_net.dfg_matrix[act1]:
v1 = heu_net.dfg_matrix[act1][act2]
v2 = heu_net.dfg_matrix[act2][act1] if act2 in heu_net.dfg_matrix and act1 in heu_net.dfg_matrix[
act2] else 0.0
tup = tuple(sorted((act1, act2)))
# added term for Heuristics Miner ++
v3 = heu_net.concurrent_activities[tup] if tup in heu_net.concurrent_activities else 0.0
dep = (v1 - v2) / (v1 + v2 + v3)
heu_net.dependency_matrix[act1][act2] = dep
if dep > dependency_thresh:
repr_value = v1 if heu_net_decoration == "frequency" else heu_net.performance_matrix[act1][act2]
heu_net.nodes[act1].add_output_connection(heu_net.nodes[act2], dep, v1, repr_value=repr_value)
heu_net.nodes[act2].add_input_connection(heu_net.nodes[act1], dep, v1, repr_value=repr_value)
return heu_net
def discover_heu_net_plus_plus(start_activities,
end_activities,
activities_occurrences,
dfg,
performance_dfg,
sojourn_time,
concurrent_activities,
parameters: Optional[Dict[Any, Any]] = None):
"""
Discovers an heuristics net using the Heuristics Miner ++ algorithm
Implements the approach described in
Burattin, Andrea, and Alessandro Sperduti. "Heuristics Miner for Time Intervals." ESANN. 2010.
https://andrea.burattin.net/public-files/publications/2010-esann-slides.pdf
Parameters
--------------
start_activities
Start activities
end_activities
End activities
activities_occurrences
Activities along with their number of occurrences
dfg
Directly-follows graph
performance_dfg
(Performance) Directly-follows graph
sojourn_time
Sojourn time for each activity
concurrent_activities
Concurrent activities
parameters
Parameters of the algorithm, including:
- Parameters.DEPENDENCY_THRESH
- Parameters.AND_MEASURE_THRESH
- Parameters.MIN_ACT_COUNT
- Parameters.MIN_DFG_OCCURRENCES
- Parameters.HEU_NET_DECORATION
Returns
--------------
heu_net
Heuristics net
"""
if parameters is None:
parameters = {}
dependency_thresh = exec_utils.get_param_value(
Parameters.DEPENDENCY_THRESH, parameters,
defaults.DEFAULT_DEPENDENCY_THRESH)
and_measure_thresh = exec_utils.get_param_value(
Parameters.AND_MEASURE_THRESH, parameters,
defaults.DEFAULT_AND_MEASURE_THRESH)
min_act_count = exec_utils.get_param_value(Parameters.MIN_ACT_COUNT,
parameters,
defaults.DEFAULT_MIN_ACT_COUNT)
min_dfg_occurrences = exec_utils.get_param_value(
Parameters.MIN_DFG_OCCURRENCES, parameters,
defaults.DEFAULT_MIN_DFG_OCCURRENCES)
heu_net_decoration = exec_utils.get_param_value(
Parameters.HEU_NET_DECORATION, parameters, "frequency")
# filter on activity and paths occurrence
activities_occurrences = {
x: y
for x, y in activities_occurrences.items() if y >= min_act_count
}
dfg = {
x: y
for x, y in dfg.items() if y >= min_dfg_occurrences
and x[0] in activities_occurrences and x[1] in activities_occurrences
}
performance_dfg = {x: y for x, y in performance_dfg.items() if x in dfg}
start_activities = {
x: y
for x, y in start_activities.items() if x in activities_occurrences
}
end_activities = {
x: y
for x, y in end_activities.items() if x in activities_occurrences
}
activities = list(activities_occurrences.keys())
if heu_net_decoration == "frequency":
heu_net = HeuristicsNet(dfg,
activities=activities,
activities_occurrences=activities_occurrences,
start_activities=start_activities,
end_activities=end_activities)
else:
heu_net = HeuristicsNet(dfg,
activities=activities,
activities_occurrences=activities_occurrences,
start_activities=start_activities,
end_activities=end_activities,
performance_dfg=performance_dfg)
heu_net.min_dfg_occurrences = min_dfg_occurrences
heu_net.sojourn_times = sojourn_time
heu_net.concurrent_activities = concurrent_activities
return calculate(heu_net, dependency_thresh, and_measure_thresh,
heu_net_decoration)
def apply_heu_dfg(dfg, activities=None, activities_occurrences=None, start_activities=None, end_activities=None,
dfg_window_2=None, freq_triples=None, performance_dfg=None, parameters=None):
"""
Discovers an Heuristics Net using Heuristics Miner
Parameters
------------
dfg
Directly-Follows Graph
activities
(If provided) list of activities of the log
activities_occurrences
(If provided) dictionary of activities occurrences
start_activities
(If provided) dictionary of start activities occurrences
end_activities
(If provided) dictionary of end activities occurrences
dfg_window_2
(If provided) DFG of window 2
freq_triples
(If provided) Frequency triples
performance_dfg
(If provided) Performance DFG
parameters
Possible parameters of the algorithm,
including:
- Parameters.ACTIVITY_KEY
- Parameters.TIMESTAMP_KEY
- Parameters.CASE_ID_KEY
- Parameters.DEPENDENCY_THRESH
- Parameters.AND_MEASURE_THRESH
- Parameters.MIN_ACT_COUNT
- Parameters.MIN_DFG_OCCURRENCES
- Parameters.DFG_PRE_CLEANING_NOISE_THRESH
- Parameters.LOOP_LENGTH_TWO_THRESH
Returns
------------
heu
Heuristics Net
"""
if parameters is None:
parameters = {}
dependency_thresh = exec_utils.get_param_value(Parameters.DEPENDENCY_THRESH, parameters,
defaults.DEFAULT_DEPENDENCY_THRESH)
and_measure_thresh = exec_utils.get_param_value(Parameters.AND_MEASURE_THRESH, parameters,
defaults.DEFAULT_AND_MEASURE_THRESH)
min_act_count = exec_utils.get_param_value(Parameters.MIN_ACT_COUNT, parameters, defaults.DEFAULT_MIN_ACT_COUNT)
min_dfg_occurrences = exec_utils.get_param_value(Parameters.MIN_DFG_OCCURRENCES, parameters,
defaults.DEFAULT_MIN_DFG_OCCURRENCES)
dfg_pre_cleaning_noise_thresh = exec_utils.get_param_value(Parameters.DFG_PRE_CLEANING_NOISE_THRESH, parameters,
defaults.DEFAULT_DFG_PRE_CLEANING_NOISE_THRESH)
loops_length_two_thresh = exec_utils.get_param_value(Parameters.LOOP_LENGTH_TWO_THRESH, parameters,
defaults.DEFAULT_LOOP_LENGTH_TWO_THRESH)
heu_net = HeuristicsNet(dfg, activities=activities, activities_occurrences=activities_occurrences,
start_activities=start_activities, end_activities=end_activities,
dfg_window_2=dfg_window_2,
freq_triples=freq_triples, performance_dfg=performance_dfg)
heu_net = calculate(heu_net, dependency_thresh=dependency_thresh, and_measure_thresh=and_measure_thresh,
min_act_count=min_act_count, min_dfg_occurrences=min_dfg_occurrences,
dfg_pre_cleaning_noise_thresh=dfg_pre_cleaning_noise_thresh,
loops_length_two_thresh=loops_length_two_thresh)
return heu_net