'numpy.float64' object cannot be interpreted as an integer

Question

Here is the code in question:

arr = ['-0.944', '0.472', '0.472']
charges = [np.float64(i) for i in arr] # [-0.944, 0.472, 0.472]
charges = np.ndarray(charges)

The error is thrown on the last step, where the list is casted to an ndarray. Assigning dtype=np.float64 in ndarray does not change the error. What is wrong with this snippet of code?

Numpy 1.14, Python 3.6.1

Answer 1

The first argument to np.ndarray is the shape, which is usually a tuple of integers.

You should not use the low-level constructor np.ndarray. Correct interface is np.array, and take it directly from the strings without a list comprehension first:

>>> arr = ['-0.944', '0.472', '0.472']
>>> np.array(arr, dtype=np.float64)
array([-0.944,  0.472,  0.472])