def test_input_diff_constraint(self):
# API and Cython object test. Replicates command line:
# ./MPP -file input.txt -thr 0.001 -att input_att1.txt -lg 20 -ug 1000 -ls 800 - us 3700 -att input_att2.txt -la 20 -ua 80 -lm 30 -um 70 -out -write BMS_patt.txt
# input on Main.cpp and verifies output with data captured from original implementation
# Similar to default but all lower constraints lowered by 10 all upper constraints raised 10.
# Significantly different results to default.
# Seq2Pat
patterns_file = self.DATA_DIR + "input.txt"
sequences = read_data(patterns_file)
seq2pat = Seq2Pat(sequences)
# Load Attributes
attribute_file = self.DATA_DIR + "input_att1.txt"
attr1_data = read_data(attribute_file)
att1 = Attribute(attr1_data)
attribute_file = self.DATA_DIR + "input_att2.txt"
attr2_data = read_data(attribute_file)
att2 = Attribute(attr2_data)
cts1 = seq2pat.add_constraint(20 <= att1.gap() <= 1000)
cts2 = seq2pat.add_constraint(800 <= att1.span() <= 3700)
cts3 = seq2pat.add_constraint(20 <= att2.average() <= 80)
cts4 = seq2pat.add_constraint(30 <= att2.median() <= 70)
test_patterns = seq2pat.get_patterns(.001)
results_file = self.DATA_DIR + "diff_constraints_results.txt"
control_patterns = read_data(results_file)
sorted_control = sort_pattern(control_patterns)
self.assertListEqual(sorted_control, test_patterns)
def test_input_no_upper_constraint(self):
# API and Cython object test. Replicates command line:
# ./MPP -file input.txt -thr 0.01 -att input_att1.txt -lg 30 -ls 900 -att input_att2.txt -la 30 -lm 40 -out -write BMS_patt.txt
# input on Main.cpp and verifies output with data captured from original implementation
# Similar to default but no upper constraints imposed
# Seq2Pat
patterns_file = self.DATA_DIR + "input.txt"
sequences = read_data(patterns_file)
seq2pat = Seq2Pat(sequences)
# Load Attributes
attribute_file = self.DATA_DIR + "input_att1.txt"
attr1_data = read_data(attribute_file)
att1 = Attribute(attr1_data)
attribute_file = self.DATA_DIR + "input_att2.txt"
attr2_data = read_data(attribute_file)
att2 = Attribute(attr2_data)
cts1 = seq2pat.add_constraint(30 <= att1.gap())
cts2 = seq2pat.add_constraint(900 <= att1.span())
cts3 = seq2pat.add_constraint(30 <= att2.average())
cts4 = seq2pat.add_constraint(40 <= att2.median())
test_patterns = seq2pat.get_patterns(.01)
results_file = self.DATA_DIR + "no_upper_constraint_results.txt"
control_patterns = read_data(results_file)
sorted_control = sort_pattern(control_patterns)
self.assertListEqual(sorted_control, test_patterns)
def test_usage(self):
# Pattern data
patterns_file = self.DATA_DIR + "input.txt"
sequences = read_data(patterns_file)
# print("Patterns: ", sequences[:5])
# Attribute data
attribute_file = self.DATA_DIR + "input_att1.txt"
attribute_1 = read_data(attribute_file)
# print("Attribute_1: ", attribute_1[:5])
# Sequential pattern finder
seq2pat = Seq2Pat(sequences)
# Constraints on attribute 1
att1 = Attribute(attribute_1)
avg_constraint = seq2pat.add_constraint(5 <= att1.average())
gap_constraint = seq2pat.add_constraint(att1.gap() <= 10)
median_constraint = seq2pat.add_constraint(10 <= att1.median() <= 15)
span_constraint = seq2pat.add_constraint(att1.span() <= 20)
# Print constraint store
# seq2pat.__str__()
seq2pat.get_patterns(min_frequency=100)
def test_median_inequality(self):
# List of sequences
sequences = [[11, 12, 13]]
# Sequential pattern finder
seq2pat = Seq2Pat(sequences)
unconstrained_result = seq2pat.get_patterns(1)
self.assertListEqual(
[[12, 13, 1], [11, 13, 1], [11, 12, 1], [11, 12, 13, 1]],
unconstrained_result)
# Attributes of sequences min median is 15 for sequence [11, 12],
# max median is 25 for sequence [12, 13]
attributes = [[10, 20, 30]]
att1 = Attribute(attributes)
# Should be empty upper and lower bounds exceed median between any two events
gap_constraint = seq2pat.add_constraint(16 <= att1.median() <= 19)
self.assertListEqual([], seq2pat.get_patterns(1))
seq2pat.remove_constraint(16 <= att1.median() <= 19)
# Should include any sequence with median between any two events with a value equal to or greater than 15
# and equal to or less than 19
gap_constraint = seq2pat.add_constraint(15 <= att1.median() <= 19)
self.assertListEqual([[11, 12, 1]], seq2pat.get_patterns(1))
seq2pat.remove_constraint(15 <= att1.median() <= 19)
# Should include any sequence with median between any two events with a value equal to or greater than 16
# and equal to or less than 20
seq2pat.add_constraint(16 <= att1.median() <= 20)
self.assertListEqual([[11, 13, 1], [11, 12, 13, 1]],
seq2pat.get_patterns(1))
seq2pat.remove_constraint(16 <= att1.median() <= 20)
# Equivalent to unconstrained span to get pattern since values of att1 fall between bounds
seq2pat.add_constraint(15 <= att1.median() <= 25)
self.assertListEqual(
[[12, 13, 1], [11, 13, 1], [11, 12, 1], [11, 12, 13, 1]],
seq2pat.get_patterns(1))
def test_usage_lb_ub(self):
# List of sequences
sequences = [[1, 2, 3], [4, 5], [1, 3, 6, 7]]
# Sequential pattern finder
seq2pat = Seq2Pat(sequences)
# Attributes of sequences
attributes = [[10, 20, 30], [40, 50], [10, 30, 60, 70]]
att1 = Attribute(attributes)
att2 = Attribute(attributes)
# Add constraints on the attributes
gap_constraint = 0 <= att1.gap() <= 10
seq2pat.add_constraint(gap_constraint)
avg_constraint = seq2pat.add_constraint(20 <= att1.average() <= 30)
median_constraint = seq2pat.add_constraint(-1 <= att1.median() <= 1000)
span_constraint = seq2pat.add_constraint(0 <= att1.span() <= 900)
span_constraint2 = seq2pat.add_constraint(-10 <= att2.span() <= 5)
# seq2pat.__str__()
self.assertEqual(gap_constraint.lower_bound, 0)
self.assertEqual(avg_constraint.lower_bound, 20)
self.assertEqual(median_constraint.lower_bound, -1)
self.assertEqual(span_constraint.lower_bound, 0)
self.assertEqual(span_constraint2.lower_bound, -10)
self.assertEqual(gap_constraint.upper_bound, 10)
self.assertEqual(avg_constraint.upper_bound, 30)
self.assertEqual(median_constraint.upper_bound, 1000)
self.assertEqual(span_constraint.upper_bound, 900)
self.assertEqual(span_constraint2.upper_bound, 5)
# Remove constraints
seq2pat.remove_constraint(gap_constraint)
seq2pat.remove_constraint(avg_constraint)
seq2pat.remove_constraint(median_constraint)
seq2pat.remove_constraint(span_constraint)
# seq2pat.__str__()
# Add again
seq2pat.add_constraint(gap_constraint)
seq2pat.add_constraint(span_constraint)
def test_usage_example_median(self):
# Seq2Pat over 3 sequences
seq2pat = Seq2Pat(sequences=[["A", "A", "B", "A", "D"],
["C", "B", "A"], ["C", "A", "C", "D"]])
# Create price attributes
price = Attribute(values=[[5, 5, 3, 8, 2], [1, 3, 3], [4, 5, 2, 1]])
# Create a time attribute
timestamp = Attribute(
values=[[1, 1, 2, 3, 3], [3, 8, 9], [2, 5, 5, 7]])
# Constraint to specify the median of prices in a pattern
med_constraint = seq2pat.add_constraint(3 <= price.median() <= 4)
# Find sequences that occur at least twice within median of prices range
patterns = seq2pat.get_patterns(min_frequency=2)
# Same solution as CMU commandline tool
self.assertListEqual([['A', 'D', 2]], patterns)
def test_seq2patfinder_default(self):
# API and Cython object test.
# Replicates command line:
# > ./MPP.exe
# -file input.txt
# -thr 0.001
# -att input_att1.txt -lg 30 -ug 900 -ls 900 - us 3600
# -att input_att2.txt -la 30 -ua 70 -lm 40 -um 60
# -out -write BMS_patt.txt
# input on Main.cpp and verifies output with data captured from original implementation
# Seq2Pat
patterns_file = self.DATA_DIR + "input.txt"
sequences = read_data(patterns_file)
seq2pat = Seq2Pat(sequences)
# Load Attributes
attribute_file = self.DATA_DIR + "input_att1.txt"
attr1_data = read_data(attribute_file)
att1 = Attribute(attr1_data)
attribute_file = self.DATA_DIR + "input_att2.txt"
attr2_data = read_data(attribute_file)
att2 = Attribute(attr2_data)
cts1 = seq2pat.add_constraint(30 <= att1.gap() <= 900)
cts2 = seq2pat.add_constraint(900 <= att1.span())
cts3 = seq2pat.add_constraint(30 <= att2.average() <= 70)
cts4 = seq2pat.add_constraint(40 <= att2.median() <= 60)
test_pf = seq2pat._get_cython_imp(-1)
self.assertListEqual([30], test_pf.lgap)
self.assertListEqual([900], test_pf.ugap)
self.assertListEqual([30], test_pf.lavr)
self.assertListEqual([70], test_pf.uavr)
self.assertListEqual([900], test_pf.lspn)
self.assertListEqual([], test_pf.uspn)
self.assertListEqual([40], test_pf.lmed)
self.assertListEqual([0], test_pf.ugapi)
self.assertListEqual([0], test_pf.lgapi)
self.assertListEqual([], test_pf.uspni)
self.assertListEqual([0], test_pf.lspni)
self.assertListEqual([1], test_pf.uavri)
self.assertListEqual([1], test_pf.lavri)
self.assertListEqual([1], test_pf.umedi)
self.assertListEqual([1], test_pf.lmedi)
self.assertListEqual([2, 0], test_pf.num_minmax)
self.assertListEqual([0, 2], test_pf.num_avr)
self.assertListEqual([0, 2], test_pf.num_med)
self.assertListEqual([0], test_pf.tot_gap)
self.assertListEqual([0], test_pf.tot_spn)
self.assertListEqual([1], test_pf.tot_avr)
self.assertEqual(161, test_pf.M)
self.assertEqual(52619, test_pf.N)
self.assertEqual(3340, test_pf.L)
self.assertListEqual([284871, 100], test_pf.max_attrs)
self.assertListEqual([0, 1], test_pf.min_attrs)
test_patterns = seq2pat.get_patterns(.001)
# Consistency sanity check
dup_patterns = seq2pat.get_patterns(.001)
self.assertListEqual(test_patterns, dup_patterns)
results_file = self.DATA_DIR + "default_results.txt"
control_patterns = read_data(results_file)
sorted_control = sort_pattern(control_patterns)
self.assertListEqual(sorted_control, test_patterns)
# Remove constraint and test
cts5 = seq2pat.remove_constraint(40 <= att2.median() <= 60)
ct6 = seq2pat.remove_constraint(30 <= att2.average() <= 70)
test_pf = seq2pat._get_cython_imp(-1)
self.assertListEqual([], test_pf.umedi)
self.assertListEqual([], test_pf.lmedi)
self.assertListEqual([], test_pf.uavr)
self.assertListEqual([], test_pf.uavri)
self.assertListEqual([0], test_pf.num_med)
one_constraint_result = seq2pat.get_patterns(.001)
results_file = self.DATA_DIR + "one_constraint_results.txt"
control_patterns = read_data(results_file)
sorted_controls = sort_pattern(control_patterns)
self.assertListEqual(sorted_controls, one_constraint_result)
