import math
import random
import time
import numpy
import os

# Comparison of speeds of various calculations

os.system('cls')

a = [random.random() for i in range(0,1000000)]

# Compute sum using in-built function
start=time.clock()
x = sum(a)
end = time.clock()

print 'Simple sum time =', (end-start)*1000.0

x=0.0

# Compute sum using FOR loop
start=time.clock()
for i in range(0,len(a)):
	x = x + a[i]
end = time.clock()

print 'FOR loop time = ', (end-start)*1000.0

x=0.0

# Compute mean with in-built functions
start=time.clock()
x = sum(a)/len(a)
end=time.clock()

print 'Simple mean time =', (end-start)*1000.0


# Compute mean using numpt
start=time.clock()
x = numpy.mean(a)
end = time.clock()

print 'Numpy mean time =', (end-start)*1000.0

x= 0.0

# Compute standard deviation using in-built functions and loops
start=time.clock()
av = sum(a)/len(a)
for i in range(0,len(a)):
	x = x + (a[i]-av)*(a[i]-av)
	sd = math.sqrt(x/len(a))
end = time.clock()

print 'Simple sdev =',sd,' Which took ',(end-start)*1000.0

# Compute standard deviation using one-line for loop
start=time.clock()
av = sum(a)/len(a)
sd = math.sqrt(sum( (a[i]-av)*(a[i]-av) for i in range(0,len(a))) / len(a) )
end=time.clock()

print 'Simple sdev 2 =',sd,' Which took ',(end-start)*1000.0


# Computer standard deviation with numpy
start = time.clock()
x = numpy.std(a)
end = time.clock()

print 'Numpy sdev =',x,' Which took ',(end-start)*1000.0

x = 0.0

# Compute median from in-built functions
start=time.clock()
asort = sorted(a)
x = asort[int(len(asort)/2.0)]
end=time.clock()

print 'Simple median =', x, 'Which took ',(end-start)*1000.0

# Compute median using numpy
start=time.clock()
x = numpy.median(a)
end=time.clock()

print 'Numpy median =', x, 'Which took', (end-start)*1000.0

l=[]

# Find values
s = time.clock()
for i in range(0,len(a)):
	if a[i] > 0.5:
		l.append(i)
e = time.clock()


print 'Finding values :',(e-s)*1000.0
	

# Extract valyes
s = time.clock()
newa = (a[i] for i in l)
e = time.clock()

print 'Extracting values :',(e-s)*1000.0
	
# Delete values
s= time.clock()
a = numpy.delete(a,l)
e = time.clock()

print 'Deleting values :',(e-s)*1000.0

print len(l)

# Summary
# Numpy is usually faster, except for computing the mean, where it is
# much slower.
# One-line for loops are x3 faster than loops over multiple lines !