➤01 产生螺旋数据

螺旋分类集合是用来测试分类器的典型试金石。在 N-arms Spiral Data 给出了产生螺旋数据集合的PYTHON程序。

• arg: 100 --arms 3
  

  ^{▲ 三条洛讯数据集合}

• arg： 200 --arms 2
  

  ^{▲ 两条螺旋线数据集合}

• arg: 100 --arms 2 --end 270
  

  ^{▲ 两条螺旋线数据集合}

➤02 PYTHON程序

#!/usr/local/bin/python
# -*- coding: gbk -*-
#============================================================
# SPIRALDATA.PY                -- by Dr. ZhuoQing 2020-11-23
#
# Note:
#============================================================

from headm import *

import argparse
import math
import numpy as np

def rotate_point(point, angle):
    """Rotate two point by an angle.
    Parameters
    ----------
    point: 2d numpy array
        The coordinate to rotate.
    angle: float
        The angle of rotation of the point, in degrees.
    Returns
    -------
    2d numpy array
        Rotated point.
    """
    rotation_matrix = np.array([[np.cos(angle), -np.sin(angle)], [np.sin(angle), np.cos(angle)]])
    rotated_point = rotation_matrix.dot(point)
    return rotated_point

def generate_spiral(samples, start, end, angle, noise):
    """Generate a spiral of points.
    Given a starting end, an end angle and a noise factor, generate a spiral of points along
    an arc.
    Parameters
    ----------
    samples: int
        Number of points to generate.
    start: float
        The starting angle of the spiral in degrees.
    end: float
        The end angle at which to rotate the points, in degrees.
    angle: float
        Angle of rotation in degrees.
    noise: float
        The noisyness of the points inside the spirals. Needs to be less than 1.
    """
    # Generate points from the square root of random data inside an uniform distribution on [0, 1).
    points = math.radians(start) + np.sqrt(np.random.rand(samples, 1)) * math.radians(end)

    # Apply a rotation to the points.
    rotated_x_axis = np.cos(points) * points + np.random.rand(samples, 1) * noise
    rotated_y_axis = np.sin(points) * points + np.random.rand(samples, 1) * noise

    # Stack the vectors inside a samples x 2 matrix.
    rotated_points = np.column_stack((rotated_x_axis, rotated_y_axis))
    return np.apply_along_axis(rotate_point, 1, rotated_points, math.radians(angle))

def main():
    parser = argparse.ArgumentParser(description='Generate n-arm spiral')
    parser.add_argument('count', type=int, help='Number of samples to generate per arm', default=2)
    parser.add_argument('--arms', type=int, help='Number of args to generate', default=2)
    parser.add_argument('--angle', type=float, help='Angle between each arm.', default=180)
    parser.add_argument('--auto-angle', type=bool, default=True,
                        help='Automatically choose the angle for the arms')
    parser.add_argument('--start', type=float, help='Start angle of the arms', default=0)
    parser.add_argument('--end', type=float, default=360,
                        help='End angle of the arms. A value of 360 corresponds \
                              to a full circle.')
    parser.add_argument('--noise', type=float, help='Noise for the arms', default=0.5)
    parser.add_argument('--filename', type=str, help='Name of the file to save the dataset',
                        default='n_arm_spiral')

    args = parser.parse_args()

    # Create a list of the angles at which to rotate the arms.
    # Either we find the angles automatically by dividing by the number of arms
    # Or we just use the angle given by the user.
    classes = np.empty((0, 3))
    angles = [((360 / args.arms) if args.auto_angle else args.angle) * i for i in range(args.arms)]

    for i, angle in enumerate(angles):
        points = generate_spiral(args.count, args.start, args.end, angle, args.noise)
        classified_points = np.hstack((points, np.full((args.count, 1), i)))
        classes = np.concatenate((classes, classified_points))

        if i == 0:      cstr = 'r'
        elif i == 1:    cstr = 'b'
        else:           cstr = 'y'

        plt.scatter(points[:,0], points[:, 1], color=cstr)

    printf(classes.shape)
    tspsave('spiral', data=classes)

    plt.xlabel("x")
    plt.ylabel("y")
    plt.grid(True)
    plt.tight_layout()
    plt.show()

#    np.savetxt(args.filename + '.csv', classes, fmt=['%10.15f', '%10.15f', '%i'], delimiter=';')

if __name__ == '__main__':
    main()

#------------------------------------------------------------
#        END OF FILE : SPIRALDATA.PY
#============================================================

  

 

  
1
  
2
  
3
  
4
  
5
  
6
  
7
  
8
  
9
  
10
  
11
  
12
  
13
  
14
  
15
  
16
  
17
  
18
  
19
  
20
  
21
  
22
  
23
  
24
  
25
  
26
  
27
  
28
  
29
  
30
  
31
  
32
  
33
  
34
  
35
  
36
  
37
  
38
  
39
  
40
  
41
  
42
  
43
  
44
  
45
  
46
  
47
  
48
  
49
  
50
  
51
  
52
  
53
  
54
  
55
  
56
  
57
  
58
  
59
  
60
  
61
  
62
  
63
  
64
  
65
  
66
  
67
  
68
  
69
  
70
  
71
  
72
  
73
  
74
  
75
  
76
  
77
  
78
  
79
  
80
  
81
  
82
  
83
  
84
  
85
  
86
  
87
  
88
  
89
  
90
  
91
  
92
  
93
  
94
  
95
  
96
  
97
  
98
  
99
  
100
  
101
  
102
  
103
  
104
  
105
  
106
  
107
  
108
  
109
  
110
 

文章来源: zhuoqing.blog.csdn.net，作者：卓晴，版权归原作者所有，如需转载，请联系作者。

原文链接：zhuoqing.blog.csdn.net/article/details/109975990