Note
Go to the end to download the full example code.
ValidationResult - Visualization#
This example demonstrates how to configure and save visualization results from Modeva to different file formats (HTML and PNG).
Installation
# To install the required package, use the following command:
# !pip install modeva
Authentication
# To get authentication, use the following command: (To get full access please replace the token to your own token)
# from modeva.utils.authenticate import authenticate
# authenticate(auth_code='eaaa4301-b140-484c-8e93-f9f633c8bacb')
Imports
from modeva import DataSet
from modeva import TestSuite
from modeva.models import MoXGBClassifier
Load and prepare data
ds = DataSet()
ds.load(name="TaiwanCredit")
ds.set_random_split()
Train models
model = MoXGBClassifier()
model.fit(ds.train_x, ds.train_y)
Generate and save plots#
Create TestSuite instances for single and multiple model analysis
ts = TestSuite(ds, model)
Limit the number of bars in bar plots#
pfi_result = ts.explain_pfi()
pfi_result.plot(n_bars=5)
List the available sub-figure names#
accuracy_results = ts.diagnose_accuracy_table()
accuracy_results.get_figure_names()
[('roc_auc', 'train'), ('precision_recall', 'train'), ('confusion_matrix', 'train'), ('roc_auc', 'test'), ('precision_recall', 'test'), ('confusion_matrix', 'test')]
Display one subplot by its name#
Note that name can be either string or tuple of string
accuracy_results.plot(name=('roc_auc', 'train'))
Save figures#
As html
pfi_result.plot_save(file_name='./image/pfi', format='html')
As png
accuracy_results.plot_save(name=('roc_auc', 'train'),
file_name='./image/compare_accuracy',
format='png')
Total running time of the script: (0 minutes 6.591 seconds)