sorend / fylearn Goto Github PK

I am getting the:

/home/user/.virtualenvs/fuzzylearning/lib/python3.5/site-packages/fylearn/frr.py:150: 
RuntimeWarning: divide by zero encountered in true_divide
  return 2 ** -((np.abs(m - S) / C) ** self.D)

warning when testing the ModifiedFuzzyPatternClassifier for my dataset.
As such the result accuracy is small, e.g. 0.1955
Any suggestions why?

ps. Thank you for your code distribution and for your time.

It is possible to implement a method to get the number of rules of the final estimator?

It will be more viable if you provide some samples to apply to this library.

Please add more info about articles you've referred to when implementing classifiers. Currently provided bibliography entries are incomplete.

For example:
fylearn/fpt.py at master:

References:
[1] Hwong, 2009.
[2] Senge and Huellemeier, 2009.
[3] Senge and Huellemeier, 2010.

Ad. 1

I believe the guy's name is Huang.

Ad. 2

There is one article from 2009 that I found. It's "Learning Pattern Tree Classifiers Using a
Co-Evolutionary Algorithm" and this doesn't sound right at all, considering that it's not about top-down fuzzy pattern trees, or so it seems.

Another article, which probably is the right one, is "Top-Down Induction of Fuzzy Pattern Trees" by those guys. More or less. Latter's name is Hüllermeier. Also, from what I see, it's from 2011, not 2009...

Please provide more info when you cite an article. This will be really helpful.

Hey, Thanks for this work in python. My query is the follwoing:
How can we implement this code to minimise my objective function, if i have multiple variables in the function that i wish to create? I am unable to define multiple variables at the moment inside the function.

say, i wish to minimise the function 2x+ 10x2 +32* z+35*z2 + x*z , with x [-5,5] and z [-5,5], How can we do that, since the TLBO generates variables based on bounds, how to use that variables in my function. Hope you can help me?

Regards,
Alan

Hello,

You have done very good work here!

If I pass an integer as random_state in local_search optimizers I get AttributeError: 'int' object has no attribute 'rand'.
In other places like in JayaOptimizer you catch this.

Since you have scikit-learn as hard dependency I would suggest to use the check_random_state utility function to create the random number generator object

http://scikit-learn.org/stable/modules/generated/sklearn.utils.check_random_state.html#sklearn.utils.check_random_state

I could send a PR for this change.

Under tests ==> test_fpt.py ==>

l = fpt.FuzzyPatternTreeClassifier()
l.fit(X, y)
score = l.score(X, y)
print("score", score)

I am not able to find the score function in main fpt.FuzzyPatternTreeClassifier() classifier as well can you please help on this

Greetings!

I'm trying to find if it is possible to use arbitrary numpy arrays as elements of the population for the search methods in the library. Or, even better, objects or dictionaries. Can you shed a light on that? Thanks in advance!

I give s sample code:

from fylearn.nfpc import FuzzyPatternClassifier
from sklearn import datasets
import numpy as np
iris = datasets.load_iris()
X = iris.data
y = iris.target
c = FuzzyPatternClassifier()
pro = c.fit(X, y).predict_proba(X)
pro_sum = pro.sum(axis=1)
print(pro_sum[np.where(pro_sum==0)].shape)

I finally get result , (12,)

I want to know how does the jaya work,and i want to use it in xgboost to optimize parameters，what can i do?thank you

Dependabot couldn't authenticate with https://pypi.python.org/simple/.

You can provide authentication details in your Dependabot dashboard by clicking into the account menu (in the top right) and selecting 'Config variables'.

View the update logs.

Do you have an example of constrained optimisation using TLBO?

Hi, Could you please provide some functional examples for working with this library?

Hey, is it possible to obtain the results of each iteration like a history?

Hey, Thanks for the response. Am still facing an issue with unpacking.