def main(args):
args_len = len(args)
if args_len in [1, 2]:
x, y = utilities.load_points_from_file(args[0])
model_funcs = get_model(x, y)
model_y = apply_funcs(x, model_funcs)
# print(model_y)
model_error = error(model_y, y)
print(model_error)
if args_len == 2:
if args[1] == '--plot':
# Print out graph
segments = len(x) // 20
for i in range(segments):
segment_x = x[i * 20:i * 20 + 20]
min_x = segment_x.min()
max_x = segment_x.max()
x_plot = np.linspace(min_x, max_x, 100)
y_plot = model_funcs[i](x_plot)
plt.plot(x_plot, y_plot)
utilities.view_data_segments(x, y)
from utilities import load_points_from_file, view_data_segments
import sys, os
from functools import reduce, partial
import numpy as np
import math
from matplotlib import pyplot as plt
# read in options
file = sys.argv[1]
plot = False
if len(sys.argv) >= 3:
plot = True if sys.argv[2] == '--plot' else False
xs, ys = load_points_from_file(file)
# returns whatever you give it :)
identity = lambda x: x
# returns y val for a point x
# for a poly a + bf(X) + cf(X)^2 + ...
def getLineVal(coeff, f, x):
y = 0
for deg in range(0, len(coeff)):
y += coeff[deg] * (f(x)**deg)
return y
class Segment():
def __init__(self, xs, ys):
return sum([c * (x**i) for i, c in enumerate(coefficients)])
def split(array, n):
if len(array) == n:
return [array]
return [array[:n]] + split(array[n:len(array)], n)
#returns the total squared error between the original data points and the estimated ones
def error(X, Y, estY):
range = abs(max(Y) - min(Y)) * abs(max(X) - min(X))
return sum([(y - y1) * (y - y1) for y, y1 in zip(Y, estY)]) / range
data = load_points_from_file(sys.argv[1])
dataX = split(data[0], segmentLength)
dataY = split(data[1], segmentLength)
def getError(X, Y, n):
poly = leastSquares(X, Y, n)
evalY = evalArray(X, poly)
return error(X, Y, evalY)
def getPolynomial(X, Y, lastError, n=2):
poly = leastSquares(X, Y, n)
evalY = evalArray(X, poly)
e = error(X, Y, evalY)
line = 0
k = 0
while k < segs[j].shape[1]:
line = line + A[j][k] * segs[j][:, k]
k = k + 1
lines = np.append(lines, line)
i = i + 20
j = j + 1
lines = lines.flatten()
plt.plot(x, lines, c='r')
plt.show()
args = sys.argv[1:]
file = args[0]
x, y = ut.load_points_from_file(file)
X = genXMatrix(x, 3) # linear, cubed and sin
r, = x.shape
i = 0
segs = []
As = []
sse = 0
while i < r: #Calculate A matrix for each line segment
X1 = genXMatrix(x[i:i + 20], 1) #Linear
A1 = leastSquares(X1, y[i:i + 20])
err1 = findError(X1, y[i:i + 20], A1)
X2 = genXMatrix(x[i:i + 20], 3) #Cubed
A2 = leastSquares(X2, y[i:i + 20])
err2 = findError(X2, y[i:i + 20], A2)
X3 = genXMatrix(x[i:i + 20], 1, 1) #Sine
A3 = leastSquares(X3, y[i:i + 20])
'across the '
'entire data set. \n o Run python lsr.py data_filename.csv --plot to calculate the total error of '
'the line approximations across the entire data set and plot the lines of best fit. \n o Run '
'python lsr.py --help for the documentation again. \n o Note that the "data_filename.csv" file '
'for data must be in the same path folder as lsr.py and must be formatted correctly.'
)
exit()
if data_filename[-4:data_filename_length] != '.csv':
print(
'Invalid first argument type. Try running "python lsr.py --help" for documentation.'
)
exit()
# Calculate the total error, don't plot the graph
adv_x, adv_y = ut1.load_points_from_file(data_filename)
ut2.plot_least_squares(adv_x, adv_y, 20, hide=False)
elif len(arguments) == 2:
# For the case: data_filename.csv --plot
if data_filename[-4:data_filename_length] != '.csv':
print(
'Invalid first argument type. Try running "python lsr.py --help" for documentation.'
)
exit()
if arguments[1] == '--plot':
# Calculate the total error, plot the graph
adv_x, adv_y = ut1.load_points_from_file(data_filename)
ut2.plot_least_squares(adv_x, adv_y, 20, hide=False, debug=False)
else:
def readFile(filename):
if (filename != None):
return load_points_from_file(filename)
else:
print("Please enter a file name")