notebook.community

Edit and run



In [12]:

    
def f1(precision,recall):
    if precision == 0.0 and recall == 0.0:
        return 0.0
    else:
        return 2 * float(precision)* float(recall) / ( precision+recall )



In [13]:

    
f1(0.31,0.76)









    Out[13]:





0.44037383177570094



In [14]:

    
f1(0.64,0.38)









    Out[14]:





0.4768627450980392



In [15]:

    
f1(0.39,0.16)









    Out[15]:





0.2269090909090909



In [16]:

    
f1(0.22,0.18)









    Out[16]:





0.19799999999999998



In [17]:

    
f1(0.11,0.39)









    Out[17]:





0.1716



In [32]:

    
f1(1,1/2)









    Out[32]:





0.6666666666666666

AP and MAP



In [ ]:

    
import numpy as np



In [ ]:

    
lst = []



In [ ]:

    
lst.append(np.array([1,2,3]))



In [ ]:

    
lst.append(np.array([4,4,4]))



In [ ]:

    
matrix = np.vstack(lst)



In [ ]:

    
matrix



In [ ]:

    
matrix.sum(axis=1)



In [ ]:

    
arr = np.array([1,2,3]).reshape(1,-1)



In [ ]:

    
np.sum(arr,axis=1)



In [ ]:

    
np.argsort(arr,axis=1)



In [ ]:

    
np.array([1,1]).shape



In [ ]:

    
original_y_true = np.array([[0.,1.],[1.,1.]]);original_y_true



In [ ]:

    
original_y_true.shape



In [ ]:

    
original_y_score = np.array([[0.,0.4],[0.9,0.1]]);original_y_score



In [ ]:

    
score_indices_top_k = np.array([[0,0],[0,0]]);score_indices_top_k



In [ ]:

    
row_indices_to_select = [i for i in range(original_y_true.shape[0])];row_indices_to_select



In [ ]:

    
column_indices_to_select = score_indices_top_k.T



In [ ]:

    
out = original_y_true[row_indices_to_select,column_indices_to_select];out



In [ ]:

    
out.shape



In [ ]:

    
arr1 = np.array([[0.5]])



In [ ]:

    
arr2 = np.array([[1.0]])



In [ ]:

    
lst = []
lst.append(arr1)
lst.append(arr2)



In [ ]:

    
np.vstack(lst)



In [ ]: