Numpy ValueError:设置一个带有序列的数组元素 [英] Numpy ValueError: setting an array element with a sequence
问题描述
当尝试将mysql数据转换为numpy数组(最终是列表)时，我得到了Numpy ValueError: setting an array element with a sequence.最初，我希望有多个字段，但是我已将代码简化为仅列出一个字段(整数)以进行故障排除.
I am getting Numpy ValueError: setting an array element with a sequence when attempting to turn mysql data into a numpy array (eventually a list). Originally I wanted to have multiple fields but I've simplified the code to only making a list with one field (integer) for troubleshooting purposes.
我是numpy的新手，所以我不确定{c1>代表什么(尽管文档说这意味着整数....好的).至于count，这似乎是指字段数.但是，该fromiter行上的某些内容仍然导致其命中异常.
I'm new to numpy so I'm not exactly sure what i4 represents (although docs say it means an integer.... ok). As for the count, that seems to refer to the number of fields. However, something on that fromiter line is still causing it to hit an exception.
import MySQLdb
import numpy
conn = MySQLdb.connect(host="localhost", user="x", passwd="x", db="x")
curs = conn.cursor()
numrows = curs.execute("select id from table")
A = numpy.fromiter(curs.fetchall(), count=numrows, dtype=('i4'))
print A
ids = A['f0']
跟踪:
A = numpy.fromiter(curs.fetchall(), count=1, dtype=('i4'))
ValueError: setting an array element with a sequence.
推荐答案
更正-要使用fetchall中的元组列表，dtype应生成结构化数组
correction - to work with a list of tuples from fetchall, the dtype should produce a structured array
查看文档，我看到fetch_all返回一个元组列表，而不是生成器.但这不是这里的问题.两者都是可迭代的.问题出在dtype.要从元组列表中创建一维数组，fromiter需要结构化的复合dtype.
Looking at the documentation I see that fetch_all returns a list of tuples, not a generator. But that isn't the issue here. Both are iterables. The problem is with the dtype. To make a 1d array from a list of tuples, fromiter requires a structured, compound, dtype.
此dtype使用1个元素的元组:
This dtype works with a 1 element tuple:
In [355]: np.fromiter([(1,)],dtype=[('f0','i4')])
Out[355]:
array([(1,)], dtype=[('f0', '<i4')])
这适用于2个字段(列)
This works with 2 fields (columns)
In [356]: np.fromiter([(1,1)],dtype=('i4,i4'))
Out[356]:
array([(1, 1)], dtype=[('f0', '<i4'), ('f1', '<i4')])
但是它们是相同的-一个普通的非结构化数组.
But these are the same - a plain, non-structured array.
np.fromiter([(1,)],dtype=('i4'))
np.fromiter([(1,)],dtype=int)
[(1,)]的处理方式与[[1]]相同，这是2d数组的输入，而不是fromiter期望的1d迭代.
[(1,)] is handled the same as [[1]], the input for a 2d array, not the 1d iterable that fromiter expects.
np.array在结构化情况下的作用与fromiter相同:
np.array works the same as fromiter for the structured cases:
np.array([(1,)],dtype=[('f0','i4')])
np.array([(1,1)],dtype=('i4,i4'))
它也可以与int(或i4)一起使用，但是结果是2d数组:
It also works with int (or i4), but the result is a 2d array:
In [366]: np.array([(1,)],dtype=('i4'))
Out[366]: array([[1]])
(早期版本)
(earlier version)
我可以通过赋予fromiter可迭代的[(1,)]来重现您的错误消息.
I can reproduce your error message by giving fromiter a [(1,)] iterable.
In [288]: np.fromiter([(1,)],dtype=int)
---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
<ipython-input-288-ba24373a9489> in <module>()
----> 1 np.fromiter([(1,)],dtype=int)
ValueError: setting an array element with a sequence.
fromiter想要1d输入，例如[1,2,3](或等效的发电机).
fromiter wants a 1d input, e.g. [1,2,3] (or the generator equivalent).
自从使用sql以来已经有一段时间了，但是我的猜测是curs.fetchall()给出了一个可迭代的元组，而不是单个数字的迭代.
It's been a while since I worked with sql, but my guess is that curs.fetchall() gives a iterable of tuples, not an iterable of single numbers.
您需要显示(打印)curs.fetchall()或list(curs.fetchall())才能查看传递给fromiter的内容.
You need to display (print) curs.fetchall() or list(curs.fetchall()) to see what what is being passed to fromiter.
为什么要使用fromiter?您尝试过np.array(curs.fetchall())吗?
Why are you using fromiter? Have you tried np.array(curs.fetchall())?
让我们尝试使用生成器表达式来更好地模拟生成元组的fetchall:
Let's try a generator expression to better simulate a fetchall that generates a tuple:
In [298]: np.fromiter((i for i in [(1,2,3)]),dtype=int)
---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
<ipython-input-298-f8fbf106b4d1> in <module>()
----> 1 np.fromiter((i for i in [(1,2,3)]),dtype=int)
ValueError: setting an array element with a sequence.
In [299]: np.array((i for i in [(1,2,3)]),dtype=int)
---------------------------------------------------------------------------
TypeError Traceback (most recent call last)
<ipython-input-299-71dd7463b539> in <module>()
----> 1 np.array((i for i in [(1,2,3)]),dtype=int)
TypeError: int() argument must be a string or a number, not 'generator'
这有效:
In [300]: np.array(list(i for i in [(1,2,3)]),dtype=int)
Out[300]: array([[1, 2, 3]])
In [301]: list(i for i in [(1,2,3)])
Out[301]: [(1, 2, 3)]
创建numpy数组的最简单方法是使用列表-它可以是数字列表，列表列表(大小均相同)或元组列表.
The simplest way to create a numpy array is with a list - it can be a list of numbers, a list of lists (all the same size) or a list of tuples.
将MySQL结果集转换为NumPy数组的最有效方法是什么?是前面关于使用fetchall和fromiter的讨论.
What's the most efficient way to convert a MySQL result set to a NumPy array? is an earlier discussion of using fetchall and fromiter.
这篇关于Numpy ValueError:设置一个带有序列的数组元素的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！
