def find_frequent(data, mis, sdc):
frequent_items = list(mis.keys())
return_list = []
# check each item to see if it's support is greater than MIS(item)
total_transactions = float( len(data) )
new_freq_items = []
for item in frequent_items:
support = float(util.actual_support(data, [[item]])) / total_transactions
if support >= mis[item]:
return_list.append( (item, mis[item]) )
return return_list
def main():
data = parsefile.parse_input_file(INPUT_FILE_NAME)
mis, sdc = parsefile.parse_param_file(PARAM_FILE_NAME)
support_dict_for_elements = util.calculate_support_for_elements(list(mis), data)
""" Step 1: Find frequent items """
frequent_items = find_frequent(data, mis, sdc)
""" Step 2: Sort frequent items in ascending order according to their MIS value """
frequent_items.sort(key=lambda x: x[1])
# remove MIS values, and just retain item ID
frequent_items = [x[0] for x in frequent_items]
# add level 1 items
final_output[1] = []
for item in frequent_items:
final_output[1].append(([[item]], util.actual_support(data, [[item]])))
for item in frequent_items:
item_mis_as_int = math.ceil(mis[item]*len(data))
transaction_subset = util.get_S_K_for_item(data, item, sdc, list(mis))
sequence_generator = util.SequenceGenerator(item, item_mis_as_int, transaction_subset, frequent_items, list(mis), support_dict_for_elements, sdc)
for i,j in sequence_generator.sequence_transaction_list:
temp = [item for sublist in i for item in sublist]
if len(temp) not in final_output:
final_output[len(temp)] = [(i, len(j))]
else:
final_output[len(temp)].append((i, len(j)))
data = util.remove_item_from_transactions(item, data)
with open(OUTPUT_FILE_NAME, 'w') as output_file:
for k in sorted(final_output.keys()):
output_file.write('\nThe number of length '+str(k)+' sequential patterns is '+ str(len(final_output[k])) + '\n\n')
for patterns in final_output[k]:
output_file.write('Pattern: '+pprint_result(str(patterns[0]))+' Count: '+str(patterns[1])+'\n')
