[numarray] mean vector of 2d matrix

[Curzio Basso]

Hi all,

I was wondering what is the best way to compute the mean vector of a
matrix. Currently I'm doing this:

>>> data.shape (117027, 8)
>>> mean = (numarray.add.reduce(data, axis=1))/(data.shape[1])

but maybe there is a faster way?

cheers

 

[beliavsky]

Hi all,

I was wondering what is the best way to compute the mean vector of a
matrix. Currently I'm doing this:

With Numeric, one can use the 'average' function to compute the average along
an axis, or the average of all elements when the axis argument is set to None.
See p317 of Martelli's "Python in a Nutshell" book. For example, the code

from Numeric import zeros,Float,average
nr = 3
nc = 2
xx = zeros([nr,nc],Float)
for i in range(nr):
for j in range(nc):
xx[i,j] = i + 10.0*j
print xx
for i in [None,0,1]: print "\n",i,average(xx,axis=i)

produces output

[[ 0. 10.]
[ 1. 11.]
[ 2. 12.]]

None 6.0

0 [ 1. 11.]

1 [ 5. 6. 7.]

 

[Curzio Basso]

With Numeric, one can use the 'average' function to compute the average along
an axis, or the average of all elements when the axis argument is set to None.
See p317 of Martelli's "Python in a Nutshell" book. For example, the code

However, it looks like in numarray no average() is provided. Now I found
out there is an average() function provided by the MA (Masked Arrays)
module, but I do not really understand why there is an average()
function for a masked array and not for a generic array...

Anyway, thanks for the tip!