Reputation: 266
I was looking to recreate Excel's trend function in Python and found the C# answer (How do I recreate an Excel formula which calls TREND() in C#?). Using this, I recreated in Python. Thought I'd post in case it was useful to anyone else.
Upvotes: 3
Views: 1793
Reputation: 51
Given function in Excel cell: =TREND(C1:C10;B1:B10;3.5)
, where C1:C10 = (1, 2, 3, 4, 5, 6, 7, 8, 9, 10)
and B1:B10 = (1, 2, 3, 4, 5, 6, 7, 8, 9, 10)
Using numpy.
import numpy as np
def trend(x, y, given_x):
# Linear regression
coefficients = np.polyfit(x, y, 1)
# Predicted y-value for a given x-value
predicted_y_for_given_x = np.polyval(coefficients, given_x)
return predicted_y_for_given_x
x = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
y = [1, 4, 4, 5, 3, 1, 0, 4, 1, 2]
given_x = 3.5
test_trend = trend(x, y, given_x)
print(test_trend) # 2.8515151515151516
Upvotes: 3
Reputation: 266
Python implementation:
def LeastSquaresValueAtX(points, knownX, xBar, x):
'''
Gets the value at a given x using the line of best fit
(least square method) to determine the equation
'''
slope = SlopeOfPoints(points, knownX, xBar)
yIntercept = YInterceptOfPoints(points, xBar, slope)
return (slope * x) + yIntercept
def SlopeOfPoints(points, knownX, xBar):
'''
Gets the slope for a set of points using the formula:
m=sum(x-avg(x)(y-avg(y))/sum(x-avg(x))**2
'''
yBar=dividend=divisor=0.0
for i in points:
yBar=yBar+i
yBar=yBar/5
for j in points:
kx = knownX.pop()
dividend+=((kx-xBar)*(j-yBar))
divisor+=((kx-xBar)**2)
return dividend / divisor
def YInterceptOfPoints(points, xBar, slope):
'''
Gets the y-intercept for a set of points using the formula:
b-avg(y)-m(avg(x))
'''
yBar =0.0
for i in points:
yBar=yBar+i
yBar=yBar/5
return yBar - (slope * xBar)
def test(knownX, t):
if t==1: return LeastSquaresValueAtX([4,13,10,22,20],[2011,2010,2009,2008,2007],2009,2012)
if t==2: return LeastSquaresValueAtX([7,20,26,29,23],knownX, 2009,2012)
if t==3: return LeastSquaresValueAtX([6,5,4,3,7], knownX, 2009,2012)
print test([2011,2010,2009,2008,2007], 1)##26.1
print test([2011,2010,2009,2008,2007], 2)##33.3
print test([2011,2010,2009,2008,2007], 3)##5.0
Upvotes: 3