Noise Detection and Label Correcting Package
Project description
NoisyDataCleaner
Python classes that identify and correct/remove noise in datasets
These models leverage on monte carlo simulation to approximate the correctness of a given label. The correction of the label builds on from the noise detection model.
Install:
pip install noisydatacleaner
Models:
-
NoiseRemover Identifies and then removes the noise from the dataset. Random Forest is used for smaller datasets as it yields better results. Whereas for larger datasets, k-Nearest Neighbors is much more efficient.
-
LabelClassificationCorrector Corrects the labels for classification datasets. Instead of only using 1 model like
NoiseRemover
, this model uses 5 different models:
models = {
'Random Forest': RandomForestClassifier(n_estimators=128),
'Extra Trees': ExtraTreesClassifier(n_estimators=128),
'Linear Discriminant': LinearDiscriminantAnalysis(),
'Logistic Regression': LogisticRegression(max_iter=128),
'Neural Network': MLPClassifier(hidden_layer_sizes=(128,64,32))
}
All of which comes from the sklearn library
Project details
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for noisydatacleaner-1.0.5-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 917fae6f86d904396281db45c0e8e7615067ab5fc251c99597ff6a2f10064bcf |
|
MD5 | f788c522664b2c747be32c25b8591347 |
|
BLAKE2b-256 | 00540d11d3f37f3362cba45daaf660b1ddcd54dc04e1651091f8033f44a74257 |