apriori_script.py

import numpy as np
from subprocess import run
import re
import os
import time
from sys import platform
import matplotlib.pyplot as plt
import seaborn as sns

from Project_Clean_data import raw, header, standardize_this
from writeapriorifile import *

X = raw

X = binarize(X)

np.savetxt("test_file.txt",X)

WriteAprioriFile(X)

if platform.startswith('linux'): #== "linux" or platform == "linux2":
    ext = ''  # Linux
    dir_sep = '/'
elif platform.startswith('darwin'): #== "darwin":
    ext = 'MAC'  # OS X
    dir_sep = '/'
elif platform.startswith('win'): #== "win32":
    ext = '.exe'  # Windows
    dir_sep = '\\'
else:
    raise NotImplementedError()

filename = "AprioriFile.txt"
minSup = 30
minConf = 60
maxRule = 4

# Run Apriori Algorithm
print('Mining for frequent itemsets by the Apriori algorithm')
status1 = run('..{0}Tools{0}apriori{1} -f"," -s{2} -v"[Sup. %S]" {3} apriori_temp1.txt'
              .format(dir_sep, ext, minSup, filename ), shell=True)

if status1.returncode != 0:
    print('An error occurred while calling apriori, a likely cause is that minSup was set to high such that no '
          'frequent itemsets were generated or spaces are included in the path to the apriori files.')
    exit()
if minConf > 0:
    print('Mining for associations by the Apriori algorithm')
    status2 = run('..{0}Tools{0}apriori{1} -tr -f"," -n{2} -c{3} -s{4} -v"[Conf. %C,Sup. %S]" {5} apriori_temp2.txt'
                  .format(dir_sep, ext, maxRule, minConf, minSup, filename ), shell=True)

    if status2.returncode != 0:
        print('An error occurred while calling apriori')
        exit()
print('Apriori analysis done, extracting results')

# Extract information from stored files apriori_temp1.txt and apriori_temp2.txt
f = open('apriori_temp1.txt', 'r')
lines = f.readlines()
f.close()
# Extract Frequent Itemsets
FrequentItemsets = [''] * len(lines)
sup = np.zeros((len(lines), 1))
for i, line in enumerate(lines):
    FrequentItemsets[i] = line[0:-1]
    sup[i] = re.findall(' [-+]?\d*\.\d+|\d+]', line)[0][1:-1]
os.remove('apriori_temp1.txt')

# Read the file
f = open('apriori_temp2.txt', 'r')
lines = f.readlines()
f.close()
# Extract Association rules
AssocRules = [''] * len(lines)
conf = np.zeros((len(lines), 1))
for i, line in enumerate(lines):
    AssocRules[i] = line[0:-1]
    conf[i] = re.findall(' [-+]?\d*\.\d+|\d+,', line)[0][1:-1]
os.remove('apriori_temp2.txt')

# sort (FrequentItemsets by support value, AssocRules by confidence value)
AssocRulesSorted = [AssocRules[item] for item in np.argsort(conf, axis=0).ravel()]
AssocRulesSorted.reverse()
FrequentItemsetsSorted = [FrequentItemsets[item] for item in np.argsort(sup, axis=0).ravel()]
FrequentItemsetsSorted.reverse()

# Print the results
time.sleep(.5)
print('\n')
print('RESULTS:\n')
print('Frequent itemsets:')
for i, item in enumerate(FrequentItemsetsSorted):
    print('Item: {0}'.format(item))
print('\n')
print('Association rules:')
for i, item in enumerate(AssocRulesSorted):
    print('Rule: {0}'.format(item))


sns.set()
pos = list(range(len(FrequentItemsetsSorted))) 
width = 0.3

ind_list = []
sup_list = []
ticks_list = ['First intercourse','Hormonal Contraceptives','Age','Num pregnacies','Num of partners']
for i in FrequentItemsetsSorted:
    ind = int(i[0])
    ind_list.append(ind)
    sup_list.append(int(i[len(i)-3:len(i)-1]))

fig, ax = plt.subplots(figsize=(10,5))
plt.bar(pos, sup_list ,width, alpha=0.5, color='g')     
#plt.bar([p + width for p in pos], , width, alpha=0.5, color='b') 

ax.set_ylabel('Support %')
ax.set_title('Frequent itemsets')
ax.set_xticks([p + width/2 for p in pos])
ax.set_xticklabels(ticks_list)

plt.xlim(min(pos)-width, max(pos)+width*2)
plt.ylim([0, 100])
plt.legend(['Percentage'], loc='upper left')

filename = "Bar_items.png"
plt.savefig(filename,dpi=600)
plt.close()

pos = list(range(len(AssocRulesSorted))) 
width = 0.2 

ind_list_cause = []
ind_list_effect = []
sup_list2 = []
conf_list = []
ticks_list = ['First intercourse','Hormonal Contraceptives','Age','Num pregnacies','Num of partners']
for i in range(len(AssocRulesSorted)):
    ind1 = int(AssocRulesSorted[i][0])
    ind2 = int(AssocRulesSorted[i][5])
    ind_list_cause.append(ind2)
    ind_list_effect.append(ind1)
    sup_list2.append(int(AssocRulesSorted[i][21:23]))
    conf_list.append(int(AssocRulesSorted[i][13:15]))

ticks_list2 = ['First intercourse->Age','Num pregnacies->Age']
fig, ax = plt.subplots(figsize=(8,5))
plt.bar(pos, sup_list2,width, alpha=0.5, color='g')     
plt.bar([p + width for p in pos], conf_list, width, alpha=0.5, color='b') 

ax.set_ylabel('Percentage')
ax.set_title('Association rules')
ax.set_xticks([p + 1.5 * width for p in pos])
ax.set_xticklabels(ticks_list2)

plt.xlim(min(pos)-width, max(pos)+width*4)
plt.ylim([0, 100])
plt.legend(['Support', 'Confidence'], loc='upper left')

filename = "Assoc_rules.png"
plt.savefig(filename,dpi=600)
plt.show()