omadson / fuzzy-c-means Goto Github PK

in the parameters:

m: represent exponent for the fuzzy partition matrix.
What represent 30% ? m=1.3 or m=30 ?

Thanks!

FuzzyC-Means is installed in the Jupyter notebook via pip install but it wont work please see the attachment.

thanks

im using python 3.11, jupyter notebook 7

pip install fuzzy-c-means

WARNING: Skipping C:\Users\Lenovo\AppData\Local\Programs\Python\Python311\Lib\site-packages\pip-23.2.1.dist-info due to invalid metadata entry 'name'
WARNING: Skipping C:\Users\Lenovo\AppData\Local\Programs\Python\Python311\Lib\site-packages\pip-23.2.1.dist-info due to invalid metadata entry 'name'
Requirement already satisfied: fuzzy-c-means in c:\users\lenovo\appdata\local\programs\python\python311\lib\site-packages (1.7.0)
Requirement already satisfied: joblib<2.0.0,>=1.2.0 in c:\users\lenovo\appdata\local\programs\python\python311\lib\site-packages (from fuzzy-c-means) (1.2.0)
Requirement already satisfied: numpy<2.0.0,>=1.21.1 in c:\users\lenovo\appdata\local\programs\python\python311\lib\site-packages (from fuzzy-c-means) (1.24.3)
Requirement already satisfied: pydantic<2.0.0,>=1.9.0 in c:\users\lenovo\appdata\local\programs\python\python311\lib\site-packages (from fuzzy-c-means) (1.10.13)
Requirement already satisfied: tabulate<0.9.0,>=0.8.9 in c:\users\lenovo\appdata\local\programs\python\python311\lib\site-packages (from fuzzy-c-means) (0.8.10)
Requirement already satisfied: tqdm<5.0.0,>=4.64.1 in c:\users\lenovo\appdata\local\programs\python\python311\lib\site-packages (from fuzzy-c-means) (4.65.0)
Requirement already satisfied: typer<0.5.0,>=0.4.0 in c:\users\lenovo\appdata\local\programs\python\python311\lib\site-packages (from fuzzy-c-means) (0.4.2)
Requirement already satisfied: typing-extensions>=4.2.0 in c:\users\lenovo\appdata\local\programs\python\python311\lib\site-packages (from pydantic<2.0.0,>=1.9.0->fuzzy-c-means) (4.6.3)
Requirement already satisfied: colorama in c:\users\lenovo\appdata\local\programs\python\python311\lib\site-packages (from tqdm<5.0.0,>=4.64.1->fuzzy-c-means) (0.4.6)
Requirement already satisfied: click<9.0.0,>=7.1.1 in c:\users\lenovo\appdata\local\programs\python\python311\lib\site-packages (from typer<0.5.0,>=0.4.0->fuzzy-c-means) (8.1.3)
WARNING: Skipping C:\Users\Lenovo\AppData\Local\Programs\Python\Python311\Lib\site-packages\pip-23.2.1.dist-info due to invalid metadata entry 'name'
WARNING: Skipping C:\Users\Lenovo\AppData\Local\Programs\Python\Python311\Lib\site-packages\pip-23.2.1.dist-info due to invalid metadata entry 'name'
WARNING: Skipping C:\Users\Lenovo\AppData\Local\Programs\Python\Python311\Lib\site-packages\pip-23.2.1.dist-info due to invalid metadata entry 'name'
WARNING: Skipping C:\Users\Lenovo\AppData\Local\Programs\Python\Python311\Lib\site-packages\pip-23.2.1.dist-info due to invalid metadata entry 'name'

The fuzzy partition coefficient (FPC) is missing? To be able to find best number of clusters.

for a given no. of clusters c, the value of partition clusters should be in the range [1/c,1] according to the research paper you linked, however here for c=2, I get:

which is clearly less than c=1/2=0.5

is there something wrong here?

I didn't find the .py file.

Hi,

I examined examples stages. I didn't see calculation auto cluster number. I guess, I have to manual enter cluster number. Is this true? Actually, I am searching something look like dbscan clustering.

I am trying to apply fuzzy c-means algorithm to cluster out my high-dimensional dataset but its not working please suggest some solution as I need to cluster data to solve class-imbalance
@omadson

I get this error when I try to access the partition_coefficient used in the provided examples. I tried updating the package but I get this error: "Could not find a version that satisfies the requirement jaxlib<0.2.0,>=0.1.57"
Could you please guide on how I can access this attribute on a Windows system?

As remembered in issue #14, we can use some metrics to validate our model. I read section 4.3.3.2. of the paper On clustering validation techniques and verified that there are some cool metrics to be implemented. I will list below the metrics that I will implement in the next few days.

• Partition coefficient (PR #15 - v1.2.2)
• Partition entropy coefficient
• Purity

Hi 👋🏻,

I am trying to reproduce the example "02 Validation". However, when I get to the point saying:

`for n_clusters, model, axe in zip(n_clusters_list, models, axes.ravel()):

# get validation metrics
pc = model.partition_coefficient
pec = model.partition_entropy_coefficient` 

I get an error saying:
'FCM' object has no attribute 'partition_coefficient'

Do I need to import something else in addition the usual packages?
It seems like it doesn't recognise the attribute partition.coefficient.

(the data X_embedded is the tSNE output as (415, 2))

Hi,

Not exactly true, but the 1.4.0 break all our debug, info and warning logs. I think that the culprit is this line. Is it possible to do something about it?

Best regards,
Jo

Perhaps I'm wrong but shouldn't a distance matrix used for clustering have small values if the samples are close and large values when they are very different? With the current implementation of the cosine similarity, we obtain +1 for the same samples and -1 for very different samples. So I think we want to change the following

to something like

 def _cosine_similarity(A: NDArray, B: NDArray) -> NDArray:
      """Compute the cosine similarity between two matrices"""
      p1 = np.sqrt(np.sum(A**2,axis=1))[:,np.newaxis]
      p2 = np.sqrt(np.sum(B**2,axis=1))[np.newaxis,:]
      return np.dot(A,B.T) / (p1*p2)

def _cosine(A: NDArray, B: NDArray) -> NDArray:
    """Compute the cosine distance between two matrices"""
    return np.abs(1 - _cosine_similarity(A, B))

and then use use _cosine instead of _cosine_similarity for computing the distance matrix.

PS: I can open a PR if required

Hey,

Really nice implementation of FCMeans. Thank you for this. I'm trying to run it on a personal server with 8 Titan Xp cards. I know they're old but I can't seem to get your model to recognize a GPU is available when I set CUDA_VISIBLE_DEVICES as an environment variable.

Am I missing something?

I get this output when I initialize the code:
WARNING:absl:No GPU/TPU found, falling back to CPU. (Set TF_CPP_MIN_LOG_LEVEL=0 and rerun for more info.)

• fuzzy-c-means 1.7.0 depends on pydantic<2.0.0 and >=1.9.0, would be nice to support > 2

I tested, there was no inf or nan in the data before hand.

I try the code below:

	from sklearn import datasets
	X, y = datasets.load_iris(return_X_y=True)
	mod = FCM(n_clusters=14, max_iter=150, m=2)
	mod.fit(X.T)

and then got this error: array must not contain infs or NaNs

I think maybe the parameter u has the nan and then norm(u - u2) wouldn't be calculated.
Hence, I have the suggestion that you can add the below code before norm(u - u2)

	if True in pd.isnull(u):
		u = u2
		break

Thanks.

Hello,
I could not find it, but is there support for custom distance metrics such a cosine similarity which is widely used in clustering in this package? Or does the FCM algorithm just use the euclidean squared distance metric?
Thanks and please let me know.

I am on Linux on a virtual machine, and i really don't understand this error, or what I have to do.
Thank you in advance.

Go to here:
https://pypi.org/project/fuzzy-c-means/

Scroll-down to where it says:

to more examples, see the examples/ folder.

https://github.com/omadson/fuzzy-c-means/tree/master/examples leads to a 404

hi, @omadson.

First of all, congratulations to your work.

I'm trying to use the CLI to fit a model on the example on the sample of README.

I just create a virtualenv and install the library. After that, I start the python REPL and type this lines:

import numpy as np

n_samples = 3000

X = np.concatenate((
    np.random.normal((-2, -2), size=(n_samples, 2)),
    np.random.normal((2, 2), size=(n_samples, 2))
))
np.savetxt("data.csv", X, delimiter=",")

After save the dataset file, I tried to use this command:

fcm fit data.csv

And I receive the following traceback:

Reading data set... Data set read without errors...
Traceback (most recent call last):
  File "/home/gustavo/.cache/pypoetry/virtualenvs/teste-fuzzy-Pa30dWmR-py3.8/bin/fcm", line 8, in <module>
    sys.exit(app())
  File "/home/gustavo/.cache/pypoetry/virtualenvs/teste-fuzzy-Pa30dWmR-py3.8/lib/python3.8/site-packages/typer/main.py", line 214, in __call__
    return get_command(self)(*args, **kwargs)
  File "/home/gustavo/.cache/pypoetry/virtualenvs/teste-fuzzy-Pa30dWmR-py3.8/lib/python3.8/site-packages/click/core.py", line 829, in __call__
    return self.main(*args, **kwargs)
  File "/home/gustavo/.cache/pypoetry/virtualenvs/teste-fuzzy-Pa30dWmR-py3.8/lib/python3.8/site-packages/click/core.py", line 782, in main
    rv = self.invoke(ctx)
  File "/home/gustavo/.cache/pypoetry/virtualenvs/teste-fuzzy-Pa30dWmR-py3.8/lib/python3.8/site-packages/click/core.py", line 1259, in invoke
    return _process_result(sub_ctx.command.invoke(sub_ctx))
  File "/home/gustavo/.cache/pypoetry/virtualenvs/teste-fuzzy-Pa30dWmR-py3.8/lib/python3.8/site-packages/click/core.py", line 1066, in invoke
    return ctx.invoke(self.callback, **ctx.params)
  File "/home/gustavo/.cache/pypoetry/virtualenvs/teste-fuzzy-Pa30dWmR-py3.8/lib/python3.8/site-packages/click/core.py", line 610, in invoke
    return callback(*args, **kwargs)
  File "/home/gustavo/.cache/pypoetry/virtualenvs/teste-fuzzy-Pa30dWmR-py3.8/lib/python3.8/site-packages/typer/main.py", line 497, in wrapper
    return callback(**use_params)  # type: ignore
  File "/home/gustavo/.cache/pypoetry/virtualenvs/teste-fuzzy-Pa30dWmR-py3.8/lib/python3.8/site-packages/fcmeans/cli.py", line 118, in fit
    model = FCM(n_clusters, max_iter, m, error, random_state)
  File "pydantic/main.py", line 397, in pydantic.main.BaseModel.__init__
TypeError: __init__() takes exactly 1 positional argument (6 given)

I'm using Python 3.8.10 on a Manjaro 5.12.19.

Can you help me on this issue?

• basic clustering
• colour quantization
• feature extraction

Currently, we use the classical setuptools with a setup.py for building the package. However, with the changes introduced by PEP517, we can now use more modern and cleaner tools, like Poetry.

These are the issues that need to be resolved:

• Create of the pyproject.toml with the necessary information about dependencies
• Document changes into the README.md to show the package building process

Hello,

I am trying to perform clustering with 300 clusters but I am getting an error. Is it possible to increase the maximum number of clusters?

Thank you,
Shahvaiz Hanif