Function not returning anything python

I am very new to Python. I am trying to write a function that does the following, and reuse the function in future parts of the code: (what the function does):

• find the cosineValue between the elements of two list
• add the values to a list and calculate the mean
• append the mean values to a list
• return the list of means

I would then like to make calculations based on the list that is returned by the above function. However, the function (i.e. knearest_similarity(tfidf_datamatrix)) does not return anything. The print commands in the second function (i.e. threshold_function())do not show anything. Can someone please have a look at the code and tell me what I am doing wrong.

def knearest_similarity(tfidf_datamatrix):
    k_nearest_cosineMean = []
    for datavector in tfidf_datamatrix:
        cosineValueSet = []
        for trainingvector in tfidf_vectorizer_trainingset:
            cosineValue = cx(datavector, trainingvector)
            cosineValueSet.append(cosineValue)
        similarityMean_of_k_nearest_neighbours = np.mean(heapq.nlargest(k_nearest_neighbours, cosineValueSet))    #the cosine similarity score of top k nearest neighbours
        k_nearest_cosineMean.append(similarityMean_of_k_nearest_neighbours)        
    print k_nearest_cosineMean
    return k_nearest_cosineMean


def threshold_function():  
    mean_cosineScore_mean = np.mean(knearest_similarity(tfidf_matrix_testset))
    std_cosineScore_mean = np.std(knearest_similarity(tfidf_matrix_testset))
    threshold = mean_cosineScore_mean - (3*std_cosineScore_mean)
    print "The Mean of the mean of cosine similarity score for a normal Behaviour:", mean_cosineScore_mean #The mean will be used for finding the threshold
    print "The standard deviation of the mean of cosine similarity score:", std_cosineScore_mean  #The standstart deviation is also used to find threshold
    print "The threshold for normal behaviour should be (Mean - 3*standard deviation):", threshold
    return threshold

EDIT

I tried defining two global variables for the functions to use (i.e. tfidf_vectorizer_trainingset and tfidf_matrix_testset).

#fitting tfidf transfrom for training data
tfidf_vectorizer_trainingset = tfidf_vectorizer.fit_transform(readfile(trainingdataDir)).toarray()

#tfidf transform the test set based on the training set
tfidf_matrix_testset = tfidf_vectorizer.transform(readfile(testingdataDir)).toarray().

However the print commands in threshold_function() appear as below:

 The Mean of the mean of cosine similarity score for a normal Behaviour: nan
The standard deviation of the mean of cosine similarity score: nan
The threshold for normal behaviour should be (Mean - 3*standard deviation): nan

EDIT2 I found that the first value in the k_nearest_cosineMean was nan. After deleting the value I managed to get valid calculations.