Python 在numpy数组中逐行查找大于阈值的值的索引
我有一个数组,如下所示。我想通过数组中的每一行查找高于阈值(例如,0.7)的值的索引Python 在numpy数组中逐行查找大于阈值的值的索引,python,numpy,threshold,numpy-ndarray,Python,Numpy,Threshold,Numpy Ndarray,我有一个数组,如下所示。我想通过数组中的每一行查找高于阈值(例如,0.7)的值的索引 items= np.array([[1. , 0.40824829, 0.03210806, 0.29488391, 0. , 0.5 , 0.32444284, 0.57735027, 0. , 0.5 ], [0.40824829, 1. , 0.57675476, 0.48154341, 0.
items= np.array([[1. , 0.40824829, 0.03210806, 0.29488391, 0. ,
0.5 , 0.32444284, 0.57735027, 0. , 0.5 ],
[0.40824829, 1. , 0.57675476, 0.48154341, 0. ,
0.81649658, 0.79471941, 0.70710678, 0.57735027, 0.40824829],
[0.03210806, 0.57675476, 1. , 0.42606683, 0. ,
0. , 0.92713363, 0.834192 , 0. , 0.73848549],
[0.29488391, 0.48154341, 0.42606683, 1. , 0. ,
0.29488391, 0.52620136, 0.51075392, 0.20851441, 0.44232587],
[0. , 0. , 0. , 0. , 0. ,
0. , 0. , 0. , 0. , 0. ],
[0.5 , 0.81649658, 0. , 0.29488391, 0. ,
1. , 0.32444284, 0.28867513, 0.70710678, 0. ],
[0.32444284, 0.79471941, 0.92713363, 0.52620136, 0. ,
0.32444284, 1. , 0.93658581, 0.22941573, 0.81110711],
[0.57735027, 0.70710678, 0.834192 , 0.51075392, 0. ,
0.28867513, 0.93658581, 1. , 0. , 0.8660254 ],
[0. , 0.57735027, 0. , 0.20851441, 0. ,
0.70710678, 0.22941573, 0. , 1. , 0. ],
[0.5 , 0.40824829, 0.73848549, 0.44232587, 0. ,
0. , 0.81110711, 0.8660254 , 0. , 1. ]])
indices_items = np.argwhere(items>= 0.7)
此(索引\u项)返回
如何按以下行获取索引?
行0->[0]行1->[0,1,5,6,7]行2->[2,6,7,9]行3->[3]行4->[]
#这应该是空列表,因为没有任何值高于阈值…获取行,使用
np列。其中,然后使用np.searchsorted
获取行数组上的区间索引,并使用这些索引拆分列数组-
In [38]: r,c = np.where(items>= 0.7)
In [39]: np.split(c,np.searchsorted(r,range(1,items.shape[0])))
Out[39]:
[array([0], dtype=int64),
array([1, 5, 6, 7], dtype=int64),
array([2, 6, 7, 9], dtype=int64),
array([3], dtype=int64),
array([], dtype=int64),
array([1, 5, 8], dtype=int64),
array([1, 2, 6, 7, 9], dtype=int64),
array([1, 2, 6, 7, 9], dtype=int64),
array([5, 8], dtype=int64),
array([2, 6, 7, 9], dtype=int64)]
就性能而言,这可能不是最优的,但如果您不真正关心它,则应该可以
indices_items = []
for l in items:
indices_items.append(np.argwhere(l >= 0.7).flatten().tolist())
indices_items
Out[5]:
[[0],
[1, 5, 6, 7],
[2, 6, 7, 9],
[3],
[],
[1, 5, 8],
[1, 2, 6, 7, 9],
[1, 2, 6, 7, 9],
[5, 8],
[2, 6, 7, 9]]
您将使用该输出做什么?您将无法将其保留为ndarray
,因为它将不再是矩形。所以它实际上不是行-列形式,而是一个链表。
indices_items = []
for l in items:
indices_items.append(np.argwhere(l >= 0.7).flatten().tolist())
indices_items
Out[5]:
[[0],
[1, 5, 6, 7],
[2, 6, 7, 9],
[3],
[],
[1, 5, 8],
[1, 2, 6, 7, 9],
[1, 2, 6, 7, 9],
[5, 8],
[2, 6, 7, 9]]