feature-selector

feature-selector copied to clipboard

Not really an issue, but just wanting to be pointed in the right direction if such a thing is possible. During the “removal all” step, it uses a lightgbm to...

windowshopr

import error ——module 'numpy.core.numerictypes' has no attribute '_typestr'

1

Hi, here is my code from feature_selector import FeatureSelector fs = FeatureSelector(data = df_ax_cla.loc[:,1:], labels = df_ax_cla.loc[:,0]) fs.identify_collinear(correlation_threshold=0.8): -------------------------------------------------------------------------- #here is error: AttributeError Traceback (most recent call last) in ---->...

chjq201410695

Hi Will, Thanks for your wonderful tool named feature-selector.I have used it to select features from my project which is a supervised learning problem successfully.But ,I have no idea to...

seualanjiang

y contains previously unseen labels: "value"

2

I have a multi-class problem where some of class values only appears in a few register, so when i run this module i got this error. fs = FeatureSelector(data =...

clappis

Zero-size array to reduction operation minimum which has no identity

1

Hello Will, Thanks for sharing your code. I tried my dataset using this code ``` from feature_selector import FeatureSelector import pandas as pd train = pd.read_csv('data/data_linkpower-Adela4.csv') train_labels = train['type_new'] print...

Adelija

There are a situation，if my data have a feature with 100% missing values, or threshold like 98% missing values, call identify_collinear() will get more features with a correlation magnitude greater...

bison31205

Add functionality to deal with datetimes in FeatureSelector. Currently the best option is to convert datetimes to seconds (for example seconds since the earliest time in the data).

WillKoehrsen

'DataFrame' object has no attribute 'append'

1

When I try to run the code for collinear features: 222 # Add to dataframe `--> 223 record_collinear = record_collinear.append(temp_df, ignore_index = True)` 'DataFrame' object has no attribute 'append'

mafaldcosta

When the data volume and feature dimensions are extremely large, the default is all cpu usage, which is not very friendly for multiple users, and the program freezes

zhangyifei1

Add link to license in readme

1

Thought it may be helpful/convenient to have a link to the license in the readme

darigovresearch