{"id":29321,"date":"2025-08-04T05:55:45","date_gmt":"2025-08-04T05:55:45","guid":{"rendered":"https:\/\/youzum.net\/tutorial-exploring-shap-iq-visualizations\/"},"modified":"2025-08-04T05:55:45","modified_gmt":"2025-08-04T05:55:45","slug":"tutorial-exploring-shap-iq-visualizations","status":"publish","type":"post","link":"https:\/\/youzum.net\/de\/tutorial-exploring-shap-iq-visualizations\/","title":{"rendered":"Tutorial: Exploring SHAP-IQ Visualizations"},"content":{"rendered":"

In this tutorial, we\u2019ll explore a range of SHAP-IQ visualizations that provide insights into how a machine learning model arrives at its predictions. These visuals help break down complex model behavior into interpretable components\u2014revealing both the individual and interactive contributions of features to a specific prediction. Check out the\u00a0Full Codes here<\/a>.<\/em><\/strong><\/p>\n

Installing the dependencies<\/strong><\/h3>\n
\n
\n
\n
\n
<\/div>\n
<\/div>\n
<\/div>\n<\/div>\n
Copy Code<\/span>Copied<\/span>Use a different Browser<\/span><\/a><\/div>\n<\/div>\n
!pip install shapiq overrides scikit-learn pandas numpy seaborn<\/code><\/pre>\n<\/div>\n<\/div>\n
\n
\n
\n
\n
<\/div>\n
<\/div>\n
<\/div>\n<\/div>\n
Copy Code<\/span>Copied<\/span>Use a different Browser<\/span><\/a><\/div>\n<\/div>\n
from sklearn.ensemble import RandomForestRegressor\nfrom sklearn.metrics import mean_squared_error, r2_score\nfrom sklearn.model_selection import train_test_split\nfrom tqdm.asyncio import tqdm\n\nimport shapiq\n\nprint(f\"shapiq version: {shapiq.__version__}\")<\/code><\/pre>\n<\/div>\n<\/div>\n
Importing the dataset<\/strong><\/h3>\n
In this tutorial, we\u2019ll use the MPG (Miles Per Gallon) dataset, which we\u2019ll load directly from the Seaborn library. This dataset contains information about various car models, including features like horsepower, weight, and origin. Check out the\u00a0Full Codes here<\/a>.<\/em><\/strong><\/p>\n
\n
\n
\n
\n
<\/div>\n
<\/div>\n
<\/div>\n<\/div>\n
Copy Code<\/span>Copied<\/span>Use a different Browser<\/span><\/a><\/div>\n<\/div>\n
import seaborn as sns\ndf = sns.load_dataset(\"mpg\")\ndf<\/code><\/pre>\n<\/div>\n<\/div>\n
Processing the dataset<\/strong><\/h3>\n
We use Label Encoding to convert the categorical column(s) into numeric format, making them suitable for model training.<\/p>\n
\n
\n
\n
\n
<\/div>\n
<\/div>\n
<\/div>\n<\/div>\n
Copy Code<\/span>Copied<\/span>Use a different Browser<\/span><\/a><\/div>\n<\/div>\n
import pandas as pd\nfrom sklearn.preprocessing import LabelEncoder\n\n# Drop rows with missing values\ndf = df.dropna()\n\n# Encoding the origin column\nle = LabelEncoder()\ndf.loc[:, \"origin\"] = le.fit_transform(df[\"origin\"])\ndf['origin'].unique()<\/code><\/pre>\n<\/div>\n<\/div>\n
\n
\n
\n
\n
<\/div>\n
<\/div>\n
<\/div>\n<\/div>\n
Copy Code<\/span>Copied<\/span>Use a different Browser<\/span><\/a><\/div>\n<\/div>\n
for i, label in enumerate(le.classes_):\n    print(f\"{label} \u2192 {i}\")<\/code><\/pre>\n<\/div>\n<\/div>\n
Splitting the data into training & test subsets<\/strong><\/h3>\n
\n
\n
\n
\n
<\/div>\n
<\/div>\n
<\/div>\n<\/div>\n
Copy Code<\/span>Copied<\/span>Use a different Browser<\/span><\/a><\/div>\n<\/div>\n
# Select features and target\nX = df.drop(columns=[\"mpg\", \"name\"])\ny = df[\"mpg\"]\n\nfeature_names = X.columns.tolist()\nx_data, y_data = X.values, y.values\n\n# Train-test split\nx_train, x_test, y_train, y_test = train_test_split(x_data, y_data, test_size=0.2, random_state=42)<\/code><\/pre>\n<\/div>\n<\/div>\n
Model Training<\/strong><\/h3>\n
We train a Random Forest Regressor with a maximum depth of 10 and 10 decision trees (n_estimators=10). A fixed random_state ensures reproducibility.<\/p>\n
\n
\n
\n
\n
<\/div>\n
<\/div>\n
<\/div>\n<\/div>\n
Copy Code<\/span>Copied<\/span>Use a different Browser<\/span><\/a><\/div>\n<\/div>\n
# Train model\nmodel = RandomForestRegressor(random_state=42, max_depth=10, n_estimators=10)\nmodel.fit(x_train, y_train)<\/code><\/pre>\n<\/div>\n<\/div>\n
Model Evaluation<\/strong><\/h3>\n
\n
\n
\n
\n
<\/div>\n
<\/div>\n
<\/div>\n<\/div>\n
Copy Code<\/span>Copied<\/span>Use a different Browser<\/span><\/a><\/div>\n<\/div>\n
# Evaluate\nmse = mean_squared_error(y_test, model.predict(x_test))\nr2 = r2_score(y_test, model.predict(x_test))\nprint(f\"Mean Squared Error: {mse:.2f}\")\nprint(f\"R2 Score: {r2:.2f}\")<\/code><\/pre>\n<\/div>\n<\/div>\n
Explaining a Local Instance<\/strong><\/h3>\n
We choose a specific test instance (with instance_id = 7) to explore how the model arrived at its prediction. We\u2019ll print the true value, predicted value, and the feature values for this instance. Check out the\u00a0Full Codes here<\/a>.<\/em><\/strong><\/p>\n
\n
\n
\n
\n
<\/div>\n
<\/div>\n
<\/div>\n<\/div>\n
Copy Code<\/span>Copied<\/span>Use a different Browser<\/span><\/a><\/div>\n<\/div>\n
# select a local instance to be explained\ninstance_id = 7\nx_explain = x_test[instance_id]\ny_true = y_test[instance_id]\ny_pred = model.predict(x_explain.reshape(1, -1))[0]\nprint(f\"Instance {instance_id}, True Value: {y_true}, Predicted Value: {y_pred}\")\nfor i, feature in enumerate(feature_names):\n    print(f\"{feature}: {x_explain[i]}\")<\/code><\/pre>\n<\/div>\n<\/div>\n
Generating Explanations for Multiple Interaction Orders<\/strong><\/h3>\n
We generate Shapley-based explanations for different interaction orders using the shapiq package. Specifically, we compute:<\/p>\n