Contents:
We first import numpy and rename it as np, then we load the data
file simple_numpy.csv using the loadtxt
function in NumPy.
import numpy as np
"""
Demonstrate some array calculations using NumPy.
"""
## Read a csv file as a matrix
X = np.loadtxt("simple_numpy.csv", delimiter=",")
The following code demonstrates a number of basic calculations that can be done on arrays using NumPy.
## Print the dimensions of the array X
print "The dimensions of X are:"
print X.shape
## Print the number of values in X that are greater than 10
print "\nThe number of entries of X that exceed 10:"
print (X > 10).sum()
## Print the proportion of values in X that are greater than 10
print "\nThe proportion of all entries of X that exceed 10:"
print (X > 10).mean()
## Print the proportion of values in each column of X that are
## greater than 10
print "\nThe proportion of entries in each column of X that exceed 10:"
print (X > 10).mean(0)
## Select the rows whose mean is greater than 5
rm = X.mean(1)
ii = np.flatnonzero(rm > 5)
Y = X[ii,:]
print "\nThe column medians of X, restricted to rows with mean exceeding 5:"
print np.median(Y, 0)
It is commonly needed to reorder the rows of an array so that the
values in one column are sorted. This can be done by using the
argsort method to obtain a vector of indices, then using these
indices to reorder the rows of the array.
## Reorder the rows of X so that the row-wise means are increasing
rm = X.mean(1)
ii = np.argsort(rm)
Y = X[ii,:]
print "\nThe first five rows of X when sorted by increasing row mean:"
print Y[0:5,:]
If we just want to sort the rows or the columns this can be done as follows:
Y = X.copy()
Y.sort(0) ## Sort within columns
Z = X.copy()
Z.sort(1) ## Sort within rows
Proportions can be calculated by taking the mean of a boolean array.
## The proportion of the rows of X in which the second column has a
## greater value than the first column
print("\nThe proportion of the rows of X in which the second column has\n"
"a greater value than the first column:")
print (X[:,1] > X[:,0]).mean()