Registry / ai-ml / mrmr-selection

mrmr-selection

library0.2.8pypypi✓ verified 84d ago

mrmr-selection implements the minimum-Redundancy-Maximum-Relevance (mRMR) algorithm for feature selection. Version 0.2.8 is current, with releases on a slow cadence. It supports classification and regression tasks, using mutual information and F-statistic to rank features.

ai-ml

pip install mrmr-selection

github ↗homepage ↗

INSTALL

IMPORT

SIG · MRMR-SELECTION

mrmr-selection

ai-mlpythonv0.2.8

harness data pending

Install & Compatibility

Where this runs

No compatibility data collected yet for this library.

Code

Verified usage

Verified import paths — ran on the pinned version, not inferred.

mrmr_classif

✓ from mrmr import mrmr_classif

✗ from mrmr_selection import mrmr_classif

Module name is 'mrmr', not 'mrmr_selection'.

mrmr_regression

✓ from mrmr import mrmr_regression

✗ import mrmr_selection

Direct import of the package fails; use 'from mrmr import'.

Select the top 2 features from the Iris dataset using mRMR classification.

import pandas as pd from sklearn.datasets import load_iris from mrmr import mrmr_classif iris = load_iris() X = pd.DataFrame(iris.data, columns=iris.feature_names) y = pd.Series(iris.target) selected_features = mrmr_classif(X=X, y=y, K=2) print(selected_features)

Debug

Known issues

breakingInput X must be a pandas DataFrame, not numpy array. Passing a numpy array silently fails or raises an error in recent versions.

fix

Convert X to a pandas DataFrame before calling mrmr_classif or mrmr_regression.

affects: >=0.2.0

deprecatedThe parameter 'n_features' was renamed to 'K' in v0.2.0. Using 'n_features' still works but will raise a DeprecationWarning.

fix

Use K=... instead of n_features=... in function calls.

affects: >=0.2.0

gotchaFor regression tasks, mrmr_regression uses F-statistic by default, not mutual information. This may not capture non-linear dependencies.

fix

If you need mutual information for regression, consider preprocessing or alternative implementations.

affects: *

gotchaThe algorithm is deterministic only if the DataFrame columns have a consistent order. Shuffling columns leads to different selections.

fix

Ensure columns are in the same order across runs or set a random_state if using tie-breaking (not yet supported).

affects: *

Errors

Common errors & fixes

ModuleNotFoundError: No module named 'mrmr_selection'

Incorrect import path; the module is named 'mrmr', not 'mrmr_selection'.

fix

pip install mrmr-selection and then use 'from mrmr import mrmr_classif'.

TypeError: X must be a pandas DataFrame

Input X is not a DataFrame (e.g., numpy array or list).

fix

Convert X to pd.DataFrame(X, columns=...) before passing to mrmr.

ValueError: The number of features to select must be positive and <= n_features

K is larger than the number of features in X or K <= 0.

fix

Set K to a positive integer less than or equal to X.shape[1].

Upgrade

Version history

0.2.8latest on PyPI · released Jun 30, 2023

Audit

Dependencies

numpyrequiredNumerical operations

pandasrequiredData handling for input DataFrames

scikit-learnrequiredMutual information and F-statistic computation

tqdmrequiredProgress bars during feature selection

Agent activity

10 hits · last 30 days

node

Resources

githubhttps://github.com/smazzanti/mrmr ↗

homepagehttps://github.com/smazzanti/mrmr ↗