blocks|key|169125|text|在您的调用中，这两个函数是相同的。|type|unstyled|depth|inlineStyleRanges|entityRanges|data|169126|不过，average可以计算加权平均值。|offset|length|style|CODE|169127|文档链接：mean和average|169128|entityMap|0|LINK|mutability|MUTABLE|url|http://docs.scipy.org/doc/numpy/reference/generated/numpy.mean.html|1|http://docs.scipy.org/doc/numpy/reference/generated/numpy.average.html^0|0|3|7|0|5|4|A|7|5|4|0|A|7|1|0^^$0|@$1|2|3|4|5|6|7|T|8|@]|9|@]|A|$]]|$1|B|3|C|5|6|7|U|8|@$D|V|E|W|F|G]]|9|@]|A|$]]|$1|H|3|I|5|6|7|X|8|@$D|Y|E|Z|F|G]|$D|10|E|11|F|G]]|9|@$D|12|E|13|1|14]|$D|15|E|16|1|17]]|A|$]]|$1|J|3|-4|5|6|7|18|8|@]|9|@]|A|$]]]|K|$L|$5|M|N|O|A|$P|Q]]|R|$5|M|N|O|A|$P|S]]]]

In your invocation, the two functions are the same.

<code>average</code> can compute a weighted average though.

Doc links: <a href="http://docs.scipy.org/doc/numpy/reference/generated/numpy.mean.html" rel="nofollow"><code>mean</code></a> and <a href="http://docs.scipy.org/doc/numpy/reference/generated/numpy.average.html" rel="nofollow"><code>average</code></a>

blocks|key|2769372|text|np.mean总是计算算术平均值，并且有一些额外的输入和输出选项(例如，使用什么数据类型，将结果放在哪里)。|type|unstyled|depth|inlineStyleRanges|offset|length|style|CODE|entityRanges|data|2769373|如果提供了weights参数，则np.average可以计算加权平均值。|2769374|entityMap^0|0|7|0|5|7|G|A|0^^$0|@$1|2|3|4|5|6|7|J|8|@$9|K|A|L|B|C]]|D|@]|E|$]]|$1|F|3|G|5|6|7|M|8|@$9|N|A|O|B|C]|$9|P|A|Q|B|C]]|D|@]|E|$]]|$1|H|3|-4|5|6|7|R|8|@]|D|@]|E|$]]]|I|$]]

<code>np.mean</code> always computes an arithmetic mean, and has some additional options for input and output (e.g. what datatypes to use, where to place the result).

<code>np.average</code> can compute a weighted average if the <code>weights</code> parameter is supplied.

blocks|key|169152|text|np.average接受一个可选的权重参数。如果没有提供，则它们是等效的。看一下源代码：Mean、Average|type|unstyled|depth|inlineStyleRanges|entityRanges|offset|length|data|169153|np.mean：|169154|try:
++++mean+=+a.mean
except+AttributeError:
++++return+_wrapit(a,+'mean',+axis,+dtype,+out)
return+mean(axis,+dtype,+out)|code-block|syntax|javascript|169155|np.average：|169156|...
if+weights+is+None+:
++++avg+=+a.mean(axis)
++++scl+=+avg.dtype.type(a.size/avg.size)
else:
++++#code+that+does+weighted+mean+here

if+returned:+#returned+is+another+optional+argument
++++scl+=+np.multiply(avg,+0)+%2B+scl
++++return+avg,+scl
else:
++++return+avg
...|169157|entityMap|0|LINK|mutability|MUTABLE|url|https://github.com/numpy/numpy/blob/master/numpy/core/fromnumeric.py|1|https://github.com/numpy/numpy/blob/master/numpy/lib/function_base.py^0|18|4|0|1D|7|1|0|0|0|0|0^^$0|@$1|2|3|4|5|6|7|Y|8|@]|9|@$A|Z|B|10|1|11]|$A|12|B|13|1|14]]|C|$]]|$1|D|3|E|5|6|7|15|8|@]|9|@]|C|$]]|$1|F|3|G|5|H|7|16|8|@]|9|@]|C|$I|J]]|$1|K|3|L|5|6|7|17|8|@]|9|@]|C|$]]|$1|M|3|N|5|H|7|18|8|@]|9|@]|C|$I|J]]|$1|O|3|-4|5|6|7|19|8|@]|9|@]|C|$]]]|P|$Q|$5|R|S|T|C|$U|V]]|W|$5|R|S|T|C|$U|X]]]]

np.average takes an optional weight parameter. If it is not supplied they are equivalent. Take a look at the source code: <a href="https://github.com/numpy/numpy/blob/master/numpy/core/fromnumeric.py" rel="noreferrer">Mean</a>, <a href="https://github.com/numpy/numpy/blob/master/numpy/lib/function_base.py" rel="noreferrer">Average</a>

np.mean:

<pre><code>try:
 mean = a.mean
except AttributeError:
 return _wrapit(a, 'mean', axis, dtype, out)
return mean(axis, dtype, out)
</code></pre>

np.average:

<pre><code>...
if weights is None :
 avg = a.mean(axis)
 scl = avg.dtype.type(a.size/avg.size)
else:
 #code that does weighted mean here

if returned: #returned is another optional argument
 scl = np.multiply(avg, 0) + scl
 return avg, scl
else:
 return avg
...
</code></pre>

blocks|key|170109|text|在某些版本的numpy中，您必须意识到另一个重要的区别：|type|unstyled|depth|inlineStyleRanges|offset|length|style|BOLD|entityRanges|data|170110|average不考虑掩码，因此计算整个数据集的平均值。|CODE|170111|mean考虑了掩码，因此仅计算未掩码的值的平均值。|170112|g+=+[1,2,3,55,66,77]
f+=+np.ma.masked_greater(g,5)

np.average(f)
Out:+34.0

np.mean(f)
Out:+2.0|code-block|syntax|javascript|170113|entityMap^0|0|B|0|0|7|0|0|4|0|0^^$0|@$1|2|3|4|5|6|7|R|8|@$9|S|A|T|B|C]]|D|@]|E|$]]|$1|F|3|G|5|6|7|U|8|@$9|V|A|W|B|H]]|D|@]|E|$]]|$1|I|3|J|5|6|7|X|8|@$9|Y|A|Z|B|H]]|D|@]|E|$]]|$1|K|3|L|5|M|7|10|8|@]|D|@]|E|$N|O]]|$1|P|3|-4|5|6|7|11|8|@]|D|@]|E|$]]]|Q|$]]

In some version of numpy there is another imporant difference that you must be aware:

<code>average</code> do not take in account masks, so compute the average over the whole set of data.

<code>mean</code> takes in account masks, so compute the mean only over unmasked values.

<pre><code>g = [1,2,3,55,66,77]
f = np.ma.masked_greater(g,5)

np.average(f)
Out: 34.0

np.mean(f)
Out: 2.0
</code></pre>

blocks|key|168154|text|除了已经提到的差异之外，还有另一个非常重要的差异，这是我刚刚通过艰难的方式发现的:与np.mean不同，np.average不允许dtype关键字，这在某些情况下对于获得正确的结果是必不可少的。我有一个从h5文件访问的非常大的单精度数组。如果我沿着轴0和1取平均值，除非我指定dtype='float64'，否则我会得到非常错误的结果|type|unstyled|depth|inlineStyleRanges|offset|length|style|CODE|entityRanges|data|168155|>T.shape
(4096,+4096,+720)
>T.dtype
dtype('<f4')

m1+=+np.average(T,+axis=(0,1))++++++++++++++++#++garbage
m2+=+np.mean(T,+axis=(0,1))+++++++++++++++++++#++the+same+garbage
m3+=+np.mean(T,+axis=(0,1),+dtype='float64')++#+correct+results|code-block|syntax|javascript|168156|不幸的是，除非你知道要找什么，否则你不一定能断定你的结果是错的。出于这个原因，我再也不会使用np.average了，但在任何大型数组上都会使用np.mean(..,+dtype='float64')。如果我想要一个加权平均值，我将使用权重向量和目标数组的乘积，然后根据需要使用np.sum或np.mean+(也具有适当的精度)来显式计算它。|168157|entityMap^0|16|7|1G|A|1T|5|2U|2|3U|F|0|0|1A|A|1Z|S|3U|6|41|7|0^^$0|@$1|2|3|4|5|6|7|O|8|@$9|P|A|Q|B|C]|$9|R|A|S|B|C]|$9|T|A|U|B|C]|$9|V|A|W|B|C]|$9|X|A|Y|B|C]]|D|@]|E|$]]|$1|F|3|G|5|H|7|Z|8|@]|D|@]|E|$I|J]]|$1|K|3|L|5|6|7|10|8|@$9|11|A|12|B|C]|$9|13|A|14|B|C]|$9|15|A|16|B|C]|$9|17|A|18|B|C]]|D|@]|E|$]]|$1|M|3|-4|5|6|7|19|8|@]|D|@]|E|$]]]|N|$]]

In addition to the differences already noted, there's another extremely important difference that I just now discovered the hard way: unlike <code>np.mean</code>, <code>np.average</code> doesn't allow the <code>dtype</code> keyword, which is essential for getting correct results in some cases. I have a very large single-precision array that is accessed from an <code>h5</code> file. If I take the mean along axes 0 and 1, I get wildly incorrect results unless I specify <code>dtype='float64'</code>:

<pre><code>&gt;T.shape
(4096, 4096, 720)
&gt;T.dtype
dtype('&lt;f4')

m1 = np.average(T, axis=(0,1)) # garbage
m2 = np.mean(T, axis=(0,1)) # the same garbage
m3 = np.mean(T, axis=(0,1), dtype='float64') # correct results
</code></pre>

Unfortunately, unless you know what to look for, you can't necessarily tell your results are wrong. I will never use <code>np.average</code> again for this reason but will always use <code>np.mean(.., dtype='float64')</code> on any large array. If I want a weighted average, I'll compute it explicitly using the product of the weight vector and the target array and then either <code>np.sum</code> or <code>np.mean</code>, as appropriate (with appropriate precision as well).

I notice that 

<pre><code>In [30]: np.mean([1, 2, 3])
Out[30]: 2.0

In [31]: np.average([1, 2, 3])
Out[31]: 2.0
</code></pre>

However, there should be some differences, since after all they are two different functions.

What are the differences between them?

np.mean() vs np.average() in Python NumPy?

我注意到In [30]: np.mean([1, 2, 3])Out[30]: 2.0In [31]: np.average([1, 2, 3])Out[31]: 2.0但是，应该有一些区别，因为它们毕竟是两个不同的功能。它们之间有什么区别？

问Python NumPy中的np.mean()与np.average()？
EN

Stack Overflow用户

社区

活动

资源

关于

腾讯云开发者

热门产品

热门推荐

更多推荐

问Python NumPy中的np.mean()与np.average()？EN