Source Code for Module TEES.Core.Split

 1  """ 
 2  Functions for dividing data into random sets. 
 3  """ 
 4   
 5  __version__ = "$Revision: 1.3 $" 
 6   
 7  import random 
 8   
 9 -def getSample(popSize, sampleFraction, seed=0): 
10      """ 
11      Generates a list of 1/0 values for defining a random sample for a list of length popSize. 
12      List elements with value 0 belong to the sample. 
13       
14      @param popSize: The length of the list from which the sample is drawn. 
15      @type popSize: int 
16      @param sampleFraction: The fraction [0,1] of the population to be included in the sample 
17      @type sampleFraction: float  
18      @param seed: int 
19      @type seed: a seed value for the Python random number generator 
20      """ 
21      random.seed(seed) 
22      sample = random.sample( xrange(popSize), int(sampleFraction*float(popSize)) ) 
23      vector = [] 
24      for i in range(popSize): 
25          if i in sample: 
26              vector.append(0) 
27          else: 
28              vector.append(1) 
29      return vector 
30   
31 -def getFolds(popSize, folds, seed=0): 
32      """ 
33      Divides the population into n folds of roughly equal size. 
34       
35      @param popSize: The length of the list from which the sample is drawn. 
36      @type popSize: int 
37      @param folds: the number of folds to divide the population into 
38      @type folds: int >= 1 
39      @param seed: int 
40      @type seed: a seed value for the Python random number generator 
41      """ 
42      sampleSize = int(float(popSize) / float(folds)) 
43      random.seed(seed) 
44       
45      vector = [] 
46      for i in range(popSize): 
47          vector.append(-1) # -1 is for items not in any fold 
48       
49      population = range(popSize) 
50      for i in range(folds): 
51          sample = random.sample(population, sampleSize) 
52          for j in sample: 
53              vector[j] = i 
54              population.remove(j) 
55      # add -1 cases roughly evenly to all folds 
56      currentFold = 0 
57      for i in range(len(vector)): 
58          if vector[i] == -1: 
59              assert(currentFold < folds-1) 
60              vector[i] = currentFold 
61              currentFold += 1 
62      return vector 
63   
64  # test program for demonstrating sampling and folds 
65  if __name__=="__main__": 
66      print "Testing 20, 0.0:" 
67      print getSample(20,0.0) 
68      print "Testing 20, 0.5:" 
69      print getSample(20,0.5) 
70      print "Folds 20 / 2:" 
71      print getFolds(20,2) 
72      print "Folds 20 / 3:" 
73      print getFolds(20,3) 
74      print "Folds 20 / 4:" 
75      print getFolds(20,4) 
76      print "Folds 20 / 20:" 
77      print getFolds(20,20) 
78