Open In Colab View Notebook on GitHub

German credit scoring [scikit-learn]#

Giskard is an open-source framework for testing all ML models, from LLMs to tabular models. Don’t hesitate to give the project a star on GitHub ⭐️ if you find it useful!

In this notebook, you’ll learn how to create comprehensive test suites for your model in a few lines of code, thanks to Giskard’s open-source Python library.

Use-case:

  • Binary classification to predict whether a person makes over 50K a year or not given their demographic variation.

  • Reference notebook

  • Dataset

Outline:

  • Detect vulnerabilities automatically with Giskard’s scan

  • Automatically generate & curate a comprehensive test suite to test your model beyond accuracy-related metrics

  • Upload your model to the Giskard Hub to:

    • Debug failing tests & diagnose issues

    • Compare models & decide which one to promote

    • Share your results & collect feedback from non-technical team members

Install dependencies#

Make sure to install the giskard

[ ]:
%pip install giskard --upgrade

Import libraries#

[2]:
from pathlib import Path
from urllib.request import urlretrieve

import pandas as pd
from sklearn.pipeline import Pipeline
from sklearn.metrics import accuracy_score
from sklearn.compose import ColumnTransformer
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler, OneHotEncoder

from giskard import Model, Dataset, scan, testing, GiskardClient, Suite

Define constants#

[3]:
# Constants
RANDOM_SEED = 0
TEST_RATIO = 0.2

DROP_FEATURES = [
    'education',
    'native-country',
    'occupation',
    'marital-status',
    'educational-num'
]

CATEGORICAL_FEATURES = [
    "workclass",
    "relationship",
    "race",
    "gender"
]

NUMERICAL_FEATURES = [
    "age",
    "fnlwgt",
    "capital-gain",
    "capital-loss",
    "hours-per-week",
]

TARGET_COLUMN = "income"

# Paths.
DATA_URL = "ftp://sys.giskard.ai/pub/unit_test_resources/wage_classification_dataset/adult.csv"
DATA_PATH = Path.home() / ".giskard" / "wage_classification_dataset" / "adult.csv"

Dataset preparation#

Load and preprocess data#

[4]:
def fetch_from_ftp(url: str, file: Path) -> None:
    """Helper to fetch data from the FTP server."""
    if not file.parent.exists():
        file.parent.mkdir(parents=True, exist_ok=True)

    if not file.exists():
        print(f"Downloading data from {url}")
        urlretrieve(url, file)

    print(f"Data was loaded!")


def download_data(**kwargs) -> pd.DataFrame:
    """Download the dataset using URL."""
    fetch_from_ftp(DATA_URL, DATA_PATH)
    _df = pd.read_csv(DATA_PATH, **kwargs)
    return _df


def preprocess_data(df: pd.DataFrame) -> pd.DataFrame:
    # Drop NaNs and columns.
    df = df.dropna()
    df = df.drop(columns=DROP_FEATURES)
    return df
[ ]:
income_df = download_data()
income_df = preprocess_data(income_df)

Train-test split#

[6]:
X_train, X_test, y_train, y_test = train_test_split(income_df.drop(columns=TARGET_COLUMN), income_df[TARGET_COLUMN],
                                                    test_size=TEST_RATIO, random_state=RANDOM_SEED)

Wrap dataset with Giskard#

To prepare for the vulnerability scan, make sure to wrap your dataset using Giskard’s Dataset class. More details here.

[7]:
raw_data = pd.concat([X_test, y_test], axis=1)
giskard_dataset = Dataset(
    df=raw_data,  # A pandas.DataFrame that contains the raw data (before all the pre-processing steps) and the actual ground truth variable (target).
    target=TARGET_COLUMN,  # Ground truth variable.
    name="salary_data",  # Optional.
    cat_columns=CATEGORICAL_FEATURES  # List of categorical columns. Optional, but is a MUST if available. Inferred automatically if not.
)

Model building#

Define preprocessing pipeline#

[8]:
preprocessor = ColumnTransformer(transformers=[
    ("num", StandardScaler(), NUMERICAL_FEATURES),
    ("cat", OneHotEncoder(handle_unknown="ignore", sparse_output=False), CATEGORICAL_FEATURES),
])

Build estimator#

[ ]:
pipeline = Pipeline(steps=[
    ("preprocessor", preprocessor),
    ("classifier", RandomForestClassifier())
])

pipeline.fit(X_train, y_train)

# Accuracy score.
train_metric = pipeline.score(X_train, y_train)
test_metric = pipeline.score(X_test, y_test)

print(f'Train accuracy: {train_metric:.2f}')
print(f'Test accuracy: {test_metric:.2f}')

Wrap model with Giskard#

To prepare for the vulnerability scan, make sure to wrap your model using Giskard’s Model class. You can choose to either wrap the prediction function (preferred option) or the model object. More details here.

[ ]:
giskard_model = Model(
    model=pipeline,  # A prediction function that encapsulates all the data pre-processing steps and that could be executed with the dataset used by the scan.
    model_type="classification",  # Either regression, classification or text_generation.
    name="salary_cls",  # Optional.
    classification_labels=pipeline.classes_,  # Their order MUST be identical to the prediction_function's output order.
    feature_names=X_train.columns  # Default: all columns of your dataset.
)

# Validate wrapped model.
wrapped_predict = giskard_model.predict(giskard_dataset)
wrapped_test_metric = accuracy_score(y_test, wrapped_predict.prediction)

print(f'Wrapped Test accuracy: {wrapped_test_metric:.2f}')

Detect vulnerabilities in your model#

Scan your model for vulnerabilities with Giskard#

Giskard’s scan allows you to detect vulnerabilities in your model automatically. These include performance biases, unrobustness, data leakage, stochasticity, underconfidence, ethical issues, and more. For detailed information about the scan feature, please refer to our scan documentation.

[ ]:
results = scan(giskard_model, giskard_dataset)
[13]:
display(results)

Generate comprehensive test suites automatically for your model#

Generate test suites from the scan#

The objects produced by the scan can be used as fixtures to generate a test suite that integrate all detected vulnerabilities. Test suites allow you to evaluate and validate your model’s performance, ensuring that it behaves as expected on a set of predefined test cases, and to identify any regressions or issues that might arise during development or updates.

[14]:
test_suite = results.generate_test_suite("My first test suite")
test_suite.run()
Executed 'Overconfidence on data slice “`hours-per-week` < 41.500”' with arguments {'model': <giskard.models.sklearn.SKLearnModel object at 0x12bccf100>, 'dataset': <giskard.datasets.base.Dataset object at 0x12bbb9de0>, 'slicing_function': <giskard.slicing.slice.QueryBasedSliceFunction object at 0x141f47550>, 'threshold': 0.5045638359329867, 'p_threshold': 0.5}:
               Test failed
               Metric: 0.51


Executed 'Underconfidence on data slice “`relationship` == "Husband"”' with arguments {'model': <giskard.models.sklearn.SKLearnModel object at 0x12bccf100>, 'dataset': <giskard.datasets.base.Dataset object at 0x12bbb9de0>, 'slicing_function': <giskard.slicing.slice.QueryBasedSliceFunction object at 0x141e86b60>, 'threshold': 0.01384993346299519, 'p_threshold': 0.95}:
               Test failed
               Metric: 0.02


Executed 'Underconfidence on data slice “`age` >= 40.500 AND `age` < 56.500”' with arguments {'model': <giskard.models.sklearn.SKLearnModel object at 0x12bccf100>, 'dataset': <giskard.datasets.base.Dataset object at 0x12bbb9de0>, 'slicing_function': <giskard.slicing.slice.QueryBasedSliceFunction object at 0x141e65060>, 'threshold': 0.01384993346299519, 'p_threshold': 0.95}:
               Test failed
               Metric: 0.02


Executed 'Underconfidence on data slice “`fnlwgt` < 128385.000 AND `fnlwgt` >= 99990.000”' with arguments {'model': <giskard.models.sklearn.SKLearnModel object at 0x12bccf100>, 'dataset': <giskard.datasets.base.Dataset object at 0x12bbb9de0>, 'slicing_function': <giskard.slicing.slice.QueryBasedSliceFunction object at 0x141e86230>, 'threshold': 0.01384993346299519, 'p_threshold': 0.95}:
               Test failed
               Metric: 0.02


Executed 'Underconfidence on data slice “`gender` == "Male"”' with arguments {'model': <giskard.models.sklearn.SKLearnModel object at 0x12bccf100>, 'dataset': <giskard.datasets.base.Dataset object at 0x12bbb9de0>, 'slicing_function': <giskard.slicing.slice.QueryBasedSliceFunction object at 0x141e870d0>, 'threshold': 0.01384993346299519, 'p_threshold': 0.95}:
               Test failed
               Metric: 0.01


Executed 'Recall on data slice “`relationship` == "Own-child"”' with arguments {'model': <giskard.models.sklearn.SKLearnModel object at 0x12bccf100>, 'dataset': <giskard.datasets.base.Dataset object at 0x12bbb9de0>, 'slicing_function': <giskard.slicing.slice.QueryBasedSliceFunction object at 0x141e5a230>, 'threshold': 0.5277777777777778}:
               Test failed
               Metric: 0.26


Executed 'Recall on data slice “`workclass` == "?"”' with arguments {'model': <giskard.models.sklearn.SKLearnModel object at 0x12bccf100>, 'dataset': <giskard.datasets.base.Dataset object at 0x12bbb9de0>, 'slicing_function': <giskard.slicing.slice.QueryBasedSliceFunction object at 0x141e70820>, 'threshold': 0.5277777777777778}:
               Test failed
               Metric: 0.33


Executed 'Recall on data slice “`relationship` == "Not-in-family"”' with arguments {'model': <giskard.models.sklearn.SKLearnModel object at 0x12bccf100>, 'dataset': <giskard.datasets.base.Dataset object at 0x12bbb9de0>, 'slicing_function': <giskard.slicing.slice.QueryBasedSliceFunction object at 0x141e589a0>, 'threshold': 0.5277777777777778}:
               Test failed
               Metric: 0.36


Executed 'Recall on data slice “`workclass` == "Self-emp-not-inc"”' with arguments {'model': <giskard.models.sklearn.SKLearnModel object at 0x12bccf100>, 'dataset': <giskard.datasets.base.Dataset object at 0x12bbb9de0>, 'slicing_function': <giskard.slicing.slice.QueryBasedSliceFunction object at 0x141e71120>, 'threshold': 0.5277777777777778}:
               Test failed
               Metric: 0.39


Executed 'Recall on data slice “`race` == "Black"”' with arguments {'model': <giskard.models.sklearn.SKLearnModel object at 0x12bccf100>, 'dataset': <giskard.datasets.base.Dataset object at 0x12bbb9de0>, 'slicing_function': <giskard.slicing.slice.QueryBasedSliceFunction object at 0x141e5a470>, 'threshold': 0.5277777777777778}:
               Test failed
               Metric: 0.39


Executed 'Recall on data slice “`relationship` == "Unmarried"”' with arguments {'model': <giskard.models.sklearn.SKLearnModel object at 0x12bccf100>, 'dataset': <giskard.datasets.base.Dataset object at 0x12bbb9de0>, 'slicing_function': <giskard.slicing.slice.QueryBasedSliceFunction object at 0x141e5a260>, 'threshold': 0.5277777777777778}:
               Test failed
               Metric: 0.41


Executed 'Recall on data slice “`gender` == "Female"”' with arguments {'model': <giskard.models.sklearn.SKLearnModel object at 0x12bccf100>, 'dataset': <giskard.datasets.base.Dataset object at 0x12bbb9de0>, 'slicing_function': <giskard.slicing.slice.QueryBasedSliceFunction object at 0x141e5a7d0>, 'threshold': 0.5277777777777778}:
               Test failed
               Metric: 0.5


[14]:
close Test suite failed. To debug your failing test and diagnose the issue, please run the Giskard hub (see documentation)
Test Overconfidence on data slice “`hours-per-week` < 41.500”
Measured Metric = 0.50771 close Failed
model 018699ef-a664-48b8-b9a2-2db999094be4
dataset salary_data
slicing_function `hours-per-week` < 41.500
threshold 0.5045638359329867
p_threshold 0.5
Test Underconfidence on data slice “`relationship` == "Husband"”
Measured Metric = 0.02269 close Failed
model 018699ef-a664-48b8-b9a2-2db999094be4
dataset salary_data
slicing_function `relationship` == "Husband"
threshold 0.01384993346299519
p_threshold 0.95
Test Underconfidence on data slice “`age` >= 40.500 AND `age` < 56.500”
Measured Metric = 0.02032 close Failed
model 018699ef-a664-48b8-b9a2-2db999094be4
dataset salary_data
slicing_function `age` >= 40.500 AND `age` < 56.500
threshold 0.01384993346299519
p_threshold 0.95
Test Underconfidence on data slice “`fnlwgt` < 128385.000 AND `fnlwgt` >= 99990.000”
Measured Metric = 0.01793 close Failed
model 018699ef-a664-48b8-b9a2-2db999094be4
dataset salary_data
slicing_function `fnlwgt` < 128385.000 AND `fnlwgt` >= 99990.000
threshold 0.01384993346299519
p_threshold 0.95
Test Underconfidence on data slice “`gender` == "Male"”
Measured Metric = 0.01496 close Failed
model 018699ef-a664-48b8-b9a2-2db999094be4
dataset salary_data
slicing_function `gender` == "Male"
threshold 0.01384993346299519
p_threshold 0.95
Test Recall on data slice “`relationship` == "Own-child"”
Measured Metric = 0.25926 close Failed
model 018699ef-a664-48b8-b9a2-2db999094be4
dataset salary_data
slicing_function `relationship` == "Own-child"
threshold 0.5277777777777778
Test Recall on data slice “`workclass` == "?"”
Measured Metric = 0.32759 close Failed
model 018699ef-a664-48b8-b9a2-2db999094be4
dataset salary_data
slicing_function `workclass` == "?"
threshold 0.5277777777777778
Test Recall on data slice “`relationship` == "Not-in-family"”
Measured Metric = 0.36364 close Failed
model 018699ef-a664-48b8-b9a2-2db999094be4
dataset salary_data
slicing_function `relationship` == "Not-in-family"
threshold 0.5277777777777778
Test Recall on data slice “`workclass` == "Self-emp-not-inc"”
Measured Metric = 0.3913 close Failed
model 018699ef-a664-48b8-b9a2-2db999094be4
dataset salary_data
slicing_function `workclass` == "Self-emp-not-inc"
threshold 0.5277777777777778
Test Recall on data slice “`race` == "Black"”
Measured Metric = 0.39167 close Failed
model 018699ef-a664-48b8-b9a2-2db999094be4
dataset salary_data
slicing_function `race` == "Black"
threshold 0.5277777777777778
Test Recall on data slice “`relationship` == "Unmarried"”
Measured Metric = 0.4127 close Failed
model 018699ef-a664-48b8-b9a2-2db999094be4
dataset salary_data
slicing_function `relationship` == "Unmarried"
threshold 0.5277777777777778
Test Recall on data slice “`gender` == "Female"”
Measured Metric = 0.50409 close Failed
model 018699ef-a664-48b8-b9a2-2db999094be4
dataset salary_data
slicing_function `gender` == "Female"
threshold 0.5277777777777778

Customize your suite by loading objects from the Giskard catalog#

The Giskard open source catalog will enable to load:

  • Tests such as metamorphic, performance, prediction & data drift, statistical tests, etc

  • Slicing functions such as detectors of toxicity, hate, emotion, etc

  • Transformation functions such as generators of typos, paraphrase, style tune, etc

To create custom tests, refer to this page.

For demo purposes, we will load a simple unit test (test_f1) that checks if the test F1 score is above the given threshold. For more examples of tests and functions, refer to the Giskard catalog.

[ ]:
test_suite.add_test(testing.test_f1(model=giskard_model, dataset=giskard_dataset, threshold=0.7)).run()

Debug and interact with your tests in the Giskard Hub#

At this point, you’ve created a test suite that is highly specific to your domain & use-case. Failing tests can be a pain to debug, which is why we encourage you to head over to the Giskard Hub.

Play around with a demo of the Giskard Hub on HuggingFace Spaces using this link.

More than just debugging tests, the Giskard Hub allows you to:

  • Compare models to decide which model to promote

  • Automatically create additional domain-specific tests through our automated model insights feature

  • Share your test results with team members and decision makers

The Giskard Hub can be deployed easily on HuggingFace Spaces. Other installation options are available in the documentation.

Here’s a sneak peek of automated model insights on a credit scoring classification model.

CleanShot 2023-09-26 at 18.38.09.png

CleanShot 2023-09-26 at 18.38.50.png

[ ]:
# Create a Giskard client after having install the Giskard server (see documentation)
api_key = "<Giskard API key>" #This can be found in the Settings tab of the Giskard hub
#hf_token = "<Your Giskard Space token>" #If the Giskard Hub is installed on HF Space, this can be found on the Settings tab of the Giskard Hub

client = GiskardClient(
    url="http://localhost:19000",  # Option 1: Use URL of your local Giskard instance.
    # url="<URL of your Giskard hub Space>",  # Option 2: Use URL of your remote HuggingFace space.
    key=api_key,
    # hf_token=hf_token  # Use this token to access a private HF space.
)

project_key = "my_project"
my_project = client.create_project(project_key, "PROJECT_NAME", "DESCRIPTION")

# Upload to the project you just created
test_suite.upload(client, project_key)

Download a test suite from the Giskard Hub#

After curating your test suites with additional tests on the Giskard Hub, you can easily download them back into your environment. This allows you to:

  • Check for regressions after training a new model

  • Automate the test suite execution in a CI/CD pipeline

  • Compare several models during the prototyping phase

[ ]:
test_suite_downloaded = Suite.download(client, project_key, suite_id=...)
test_suite_downloaded.run()