Membership Gaussmf

[scikit-fuzzy]

Membership Gaussmf

[JDWarner]

@NetzerKathrin I can't reproduce this. The operations are carried out floating-point, and even if I go out of my way to give the function integer inputs with a narrow sigma the result is still floating point. This

import numpy as np
import skfuzzy as fuzz

print(fuzz.gaussmf(np.asarray([0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10], np.int16), 5, 1))

returns

[  1.38879439e-11   1.12535175e-07   1.23409804e-04   1.83156389e-02
   3.67879441e-01   1.00000000e+00   3.67879441e-01   1.83156389e-02
   1.23409804e-04   1.12535175e-07   1.38879439e-11]

Note that with a relatively narrow Gaussian (small sigma) these do get quite small rather quickly. This behavior depends on NumPy, so it should be platform-independent.

Perhaps your display settings for NumPy arrays round instead of displaying scientific notation?

[NetzerKathrin]

Thanks for your answer. I got the exception that told me my antecedents would be too sparse. After debugging it and looking into the terms of the antecedents I defined I noticed that at some point the values actually turned zero (0.0). Perhaps you are right and it is because of some settings. I worked around the exception by replacing these 0.0 by the smallest ...e-... value in the terms[..].mf Am 22.04.2017 um 13:37 schrieb Josh Warner:

…

@NetzerKathrin <https://github.com/NetzerKathrin> I can't reproduce this. The operations are carried out floating-point, and even if I go out of my way to give the function integer inputs with a narrow sigma the result is still floating point. This import numpyas np import skfuzzyas fuzz print(fuzz.gaussmf(np.asarray([0,1,2,3,4,5,6,7,8,9,10], np.int16),5,1)) returns [1.38879439e-11 1.12535175e-07 1.23409804e-04 1.83156389e-02 3.67879441e-01 1.00000000e+00 3.67879441e-01 1.83156389e-02 1.23409804e-04 1.12535175e-07 1.38879439e-11] Note that with a relatively narrow Gaussian (small sigma) these do get quite small rather quickly. This behavior depends on NumPy, so it should be platform-independent. Perhaps your display settings for NumPy arrays round instead of displaying scientific notation? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#132 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ATBvLUagyKX7IruMWXUKGhLUkFyeXOJcks5ryeZ-gaJpZM4NFBZo>.

[JDWarner]

Closing as cannot reproduce, but feel free to reopen with a brief example so I can track it down if you believe this remains an issue.