atts 0.0.8

Train_test splitter with adversarial validation

ATTS (Auto Train Test Splitter) splits the given dataframe to train and test taking into account the Concept Drift. It does this with xgboost running in the background.

Calculates roc_auc scores for the given test size values. Creates a dataframe with these values. Provides easier change visibility with chart.

The code is Python 3

Installation

Fast install:

pip install atts

For a manual install get this package:

wget https://github.com/alikula314/att_split/atts/archive/master.zip
unzip master.zip
rm master.zip
cd atts-master

Install the package:

python setup.py install

Example on Red Wine Quality Dataset

pip install atts
from atts.auto_test_splitter import atts

# run module
k = ATTS(df, "quality" , 0.05, 0.50, 0.05)

# get auc_score - test_size dataframe
k.test_sizes_df()

#get the plot
k.test_sizes_plot(800,400)