adds the functions:

averageDuration -- computers the average duration for the specific part
in a course.
splitSequence -- split and rebuild the construction of data in order to
extract and analysis arguments more easily.
averageLengthForPart -- get the average numbers of parts seen in a
course for all users

2015-Hubble-UserProfile/BukaiLinNew/dataProcessing.py

+# -*- coding: UTF-8 -*-
+'''
+Created on 2016��6��3��
+
+@author: E15E888T
+'''
+from dataManager.scriptDatabaseManager2 import ScriptDatabaseManager
+from dataManager.databaseManagerData2 import DatabaseManager
+from dataManager.scriptDatabaseStatistic import ScriptDatabaseStatistic
+
+import math
+import os
+import pickle
+from asyncore import read
+from lib2to3.pgen2.tokenize import Special
+from pattern.metrics import duration
+
+databaseManager = DatabaseManager()
+scriptDatabaseManager = ScriptDatabaseManager()
+
+class DataProcessing:
+
+    def averageDuration(self, idPart, sequence):
+        '''
+        get the average duration of the special part in a special course
+        '''
+        sum, ave, n = 0, 0, 0
+              
+        sequence = self.splitSequence(sequence, 'idPart+duration', False)
+         
+        for i in range(0,len(sequence)):
+            for j in range(0,len(sequence[i])):
+                if sequence[i][j][0] == idPart:
+                    sum = sum + sequence[i][j][1]
+                    n = n + 1
+        ave = float(sum)/float(n)
+        return ave   
+         
+    
+    def splitSequence(self, sequence, argument, needToSave):
+        '''
+        get the different arguments of the users for a specific course
+                
+        return: a list [idPart, idPart, ... ,idPart], [idSession, idSession, ... ,idSession]
+            
+        [duration, duration, ... ,duration] or [((idPart, duration),(idPart, duration)...)...((idPart, duration),(idPart, duration)...)]
+        '''
+        list = []
+        list.append(sequence)
+            
+        (idCourse, firstSequence) = list[0]
+        userid = []; partid = []; sessionid = []; duration = []; 
+        secondSequence = [] ; thirdSequence = []; fourthSequence = []; finalSequence = []
+        
+        for i in range(0,len(firstSequence)):
+            userid.append(firstSequence[i][0])
+            for j in range(0,len(firstSequence[i][1])):
+                partid.append(firstSequence[i][1][j][0])
+                sessionid.append(firstSequence[i][1][j][1])
+                duration.append(firstSequence[i][1][j][2])
+                a = tuple((firstSequence[i][1][j][0],firstSequence[i][1][j][2]))
+                secondSequence.append(a)
+            if i != 0:
+                del secondSequence[0:len(firstSequence[i-1][1])]
+            b = tuple(secondSequence)
+            if argument == 'idUser+idPart+duration':
+                thirdSequence.insert(2*i, firstSequence[i][0])
+                thirdSequence.append(b)
+                if i != 0:
+                    del thirdSequence[0:2]
+                c = tuple(thirdSequence)
+                fourthSequence.append(c)
+            else:
+                thirdSequence.append(b)
+        
+        if argument == 'idUser':
+            finalSequence = userid
+        elif argument == 'idPart':
+            finalSequence = partid
+        elif argument == 'idSession':
+            finalSequence =  sessionid
+        elif argument == 'duration':
+            finalSequence =  duration
+        elif argument == 'idPart+duration':
+            finalSequence =  thirdSequence
+        elif argument == 'idUser+idPart+duration':
+            finalSequence =  fourthSequence
+        else:
+            print ("Wrong input type")
+            
+        if needToSave:
+            fileName = "filterSequences"+str(idCourse)+".p"
+            pickle.dump(finalSequence, open(os.path.join('filtersequences', fileName), "wb"))
+         
+        return finalSequence
+        
+
+    def averageLengthForPart(self, idCourse):
+        '''
+        get the average length of part in a course
+        '''
+        scriptDatabaseManager = ScriptDatabaseManager()
+        a = scriptDatabaseManager.getSequencesSucceedPremium(idCourse)
+        
+        idpart = self.splitSequence(a, 'idPart', False)
+        iduser = self.splitSequence(a, 'idUser', False)
+        
+        ave = float(len(idpart))/float(len(iduser))
+        return ave   
+        
+