# # Copyright (c) 2018 James Hume (www.jeh-tech.com). All rights reserved. # # Redistribution and use in source and binary forms, with or without # modification, are permitted provided that the following conditions # are met: # 1. Redistributions of source code must retain the above copyright # notice, this list of conditions and the following disclaimer. # 2. Redistributions in binary form must reproduce the above copyright # notice, this list of conditions and the following disclaimer in the # documentation and/or other materials provided with the distribution. # 3. All advertising materials mentioning features or use of this software # must display the following acknowledgement: # This product includes software developed by James Hume (www.jeh-tech.com) # 4. Neither the name "James Hume" or website "www.jeh-tech.com" # may be used to endorse or promote products derived from this software # without specific prior written permission. # # THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND # ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE # IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE # ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE # FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL # DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS # OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) # HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT # LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY # OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF # SUCH DAMAGE. # import matplotlib.pyplot as pl import numpy as np import matplotlib.animation as animation import math import numpy as np ## ## Project a vector onto the line defined by the vector v ## x - vector (numpy array) to be projected onto the line L ## v - vector (numpy array) defining the line L: L = {a * v} def proj_x_onto_L(x, v): return (x.dot(v) / np.power(np.linalg.norm(v), 2)) * v ## ## Plot a 2D vector starting at an origin ## vec - vector (array or list) to plot, starting at origin ## origin - origin point of vector def plot_vec(vec, origin, *args, **kwargs): if origin is not None: return ax.plot([origin[0], origin[0] + vec[0]], [origin[1], origin[1] + vec[1]], *args, **kwargs)[0] else: return ax.plot([0, vec[0]], [0, vec[1]], *args, **kwargs)[0] ## ## Plot 2D line plane ## ## Get the limits of the x-axis and plot a line that spans the entire x-axis to make ## the line look "infinite" ## ## vec - Vector (array or list) defining the line vector m if L = {a * m + b} ## yintercept - This is b if L = {a * m + b} def plot_line_plane(vec, yintercept, *args, **kwargs): xmin, xmax = ax.get_xlim() m = vec[1]/vec[0] ymin = xmin * m + yintercept ymax = xmax * m + yintercept return ax.plot([xmin, xmax], [ymin, ymax], *args, **kwargs)[0] def annotate_line(line, annotate_at_x, annotate_txt, xytext): x1, x2 = line.get_xdata() y1, y2 = line.get_ydata() m = (y2 - y1) / (x2 - x1) yintercept = y1 - m*x1 pl.annotate( annotate_txt , xy=(annotate_at_x, annotate_at_x * m + yintercept) , xytext=xytext , textcoords='offset points' , fontsize='medium' , arrowprops=dict(shrink=0.05, connectionstyle="arc3,rad=0.1", fc=line.get_color()) ) ## ## Setup figure and axis so that graph axis at (0,0) and so that x/y limits ## leave enough room to make figure look "good" fig, ax = pl.subplots() ax.set_xlim([-1.5, 2.5]) ax.set_ylim([-0.5, 2.5]) ax.spines['left'].set_position('zero') ax.spines['bottom'].set_position('zero') ax.spines['top'].set_color('none') ax.spines['right'].set_color('none') ax.set_title('Projections calculated using data transformations to make LSS pass\nthrough origin. After projection onto LSS through origin everything\ntranslated back') ## ## Define noisy data points data_points = [ np.array([-1, 0]), np.array([0, 1]), np.array([1, 2]), np.array([2, 1]) ] ## ## A color for each noisey data point colours = ['red', 'green', 'blue', 'purple'] ## ## Define the least squares solution lss_vec = np.array([1., 2./5.]) lss_yintercept = np.array([0, 4./5.]) ## ## Apply the translation that would bring the LSS line to the origin to all the data points for i in range(len(data_points)): data_points[i] = data_points[i] - lss_yintercept ## ## Build projection of each data point vector onto LSS projections = [] for dp in data_points: projections.append(proj_x_onto_L(dp, lss_vec)) ## ## Build the orthogonal line from each point to its projection on the LSS line orthogs = [] for dp, p in zip(data_points, projections): orthogs.append(p - dp) ## ## Apply the inverse of the translation that would bring the LSS line to the origin to ## all the data points, to put them back where they should be for i in range(len(data_points)): data_points[i] = data_points[i] + lss_yintercept ## ## Plot the data points for i, (dp, c) in enumerate(zip(data_points, colours)): ax.plot(dp[0], dp[1], 'x', color=c) ax.text(dp[0], dp[1] + 0.03, '$\\vec{{d_{}}}$'.format(i), color=c) ## ## Plot the least squares solution line lss_line = plot_line_plane(lss_vec, lss_yintercept[1], color='grey') ## ## Plot the projections - and here we see that the projections are onto the line ## passing through the origin, because to project onto a vector, it must pass through ## the origin because if L = c \vec{v}, for all real c's, when c is zero, it passes ## through origin. for dp, p, orth, c in zip(data_points, projections, orthogs, colours): print(orth) plot_vec(orth, dp, color=c, alpha=0.5) annotate_line(lss_line, -1.25, r'Least squares solution', (-40, 35)) fig.show() pl.show()