I'm new to python and numpy and I wanted to create an array with mixed data types but I get this error :
>>> m = np.zeros((5,10))
>>> m[0,0] = 'e'
Traceback (most recent call last):
File "<pyshell#17>", line 1, in <module>
m[0,0] = 'e'
ValueError: could not convert string to float: e
I'm using numpy beacuse it's easier to implement bigger sized arrays and edit them. I tried to change the data type of the array but I couldn't find any suitable one.
If you really, really want an array with mixed data types, use dtype = object.
m = np.zeros((5,10), dtype = object)
m[0,0] = 'e'
m
Out[105]:
array([['e', 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0]], dtype=object)
Now, you probably don't want that really, since object arrays are fairly useless and most numpy functions don't work with them. If you want to import a database into an array, for instance, you will want to use a Structured Array. This will let you mix strings and numbers in such a way that you can still do something with them via numpy
If you really just want a way to deal with big databases using an efficient indexing scheme, you probably don't want numpy at all, but but instead want pandas or even dask. numpy is optimized for calculation, not storage.
dtypeparameter. See this