sklearn-docs/GradientBoostingClassifier

1import gradio as gr2import numpy as np3import pandas as pd4import matplotlib5import matplotlib.pyplot as plt6 7from datasets import load_dataset8from sklearn.ensemble import GradientBoostingClassifier9from sklearn.model_selection import train_test_split10from sklearn.metrics import accuracy_score, confusion_matrix11 12matplotlib.use('Agg')13 14################################################################################15# SUGGESTED_DATASETS: These must actually exist on huggingface.co/datasets16#17# "scikit-learn/iris" -> A small, classic Iris dataset with a "train" split18# "uci/wine"          -> Another small dataset with a "train" split19# "SKIP/ENTER_CUSTOM" -> Placeholder to let the user enter a custom dataset ID20################################################################################21SUGGESTED_DATASETS = [22    "scikit-learn/iris",23    "uci/wine",24    "SKIP/ENTER_CUSTOM"25]26 27def update_columns(dataset_id, custom_dataset_id):28    """29    After the user chooses a dataset from the dropdown or enters their own,30    this function loads the dataset's "train" split, converts it to a DataFrame,31    and returns the columns. These columns are used to populate the Label and32    Feature selectors in the UI.33    """34    if dataset_id != "SKIP/ENTER_CUSTOM":35        final_id = dataset_id36    else:37        final_id = custom_dataset_id.strip()38 39    try:40        ds = load_dataset(final_id, split="train")41        df = pd.DataFrame(ds)42        cols = df.columns.tolist()43 44        message = (45            f"**Loaded dataset**: `{final_id}`\n\n"46            f"**Columns found**: {cols}"47        )48        return (49            gr.update(choices=cols, value=None),   # label_col dropdown50            gr.update(choices=cols, value=[]),     # feature_cols checkbox group51            message52        )53    except Exception as e:54        err_msg = f"**Error loading** `{final_id}`: {e}"55        return (56            gr.update(choices=[], value=None),57            gr.update(choices=[], value=[]),58            err_msg59        )60 61def train_model(dataset_id, custom_dataset_id, label_column, feature_columns,62                learning_rate, n_estimators, max_depth, test_size):63    """64    1. Decide which dataset ID to load (from dropdown or custom).65    2. Load that dataset's 'train' split, turn into DataFrame, extract X (features) and y (label).66    3. Train a GradientBoostingClassifier on X_train, y_train.67    4. Compute accuracy and confusion matrix on X_test, y_test.68    5. Plot and return feature importances + confusion matrix heatmap + textual summary.69    """70    # Resolve final dataset ID71    if dataset_id != "SKIP/ENTER_CUSTOM":72        final_id = dataset_id73    else:74        final_id = custom_dataset_id.strip()75 76    # Load dataset -> df77    ds = load_dataset(final_id, split="train")78    df = pd.DataFrame(ds)79 80    # Validate columns81    if label_column not in df.columns:82        raise ValueError(f"Label column '{label_column}' not found in dataset columns.")83    for fc in feature_columns:84        if fc not in df.columns:85            raise ValueError(f"Feature column '{fc}' not found in dataset columns.")86 87    # Convert to NumPy arrays88    X = df[feature_columns].values89    y = df[label_column].values90 91    # Train/test split92    X_train, X_test, y_train, y_test = train_test_split(93        X, y, test_size=test_size, random_state=4294    )95 96    # Instantiate and train GradientBoostingClassifier97    clf = GradientBoostingClassifier(98        learning_rate=learning_rate,99        n_estimators=int(n_estimators),100        max_depth=int(max_depth),101        random_state=42102    )103    clf.fit(X_train, y_train)104 105    # Evaluate106    y_pred = clf.predict(X_test)107    accuracy = accuracy_score(y_test, y_pred)108    cm = confusion_matrix(y_test, y_pred)109 110    # Create Matplotlib figure with feature importances + confusion matrix111    fig, axs = plt.subplots(1, 2, figsize=(10, 4))112 113    # Subplot 1: Feature Importances114    importances = clf.feature_importances_115    axs[0].barh(range(len(feature_columns)), importances, color='skyblue')116    axs[0].set_yticks(range(len(feature_columns)))117    axs[0].set_yticklabels(feature_columns)118    axs[0].set_xlabel("Importance")119    axs[0].set_title("Feature Importances")120 121    # Subplot 2: Confusion Matrix Heatmap122    im = axs[1].imshow(cm, interpolation='nearest', cmap=plt.cm.Blues)123    axs[1].set_title("Confusion Matrix")124    plt.colorbar(im, ax=axs[1])125    axs[1].set_xlabel("Predicted")126    axs[1].set_ylabel("True")127 128    # Optionally annotate each cell with numeric counts129    thresh = cm.max() / 2.0130    for i in range(cm.shape[0]):131        for j in range(cm.shape[1]):132            color = "white" if cm[i, j] > thresh else "black"133            axs[1].text(j, i, str(cm[i, j]), ha="center", va="center", color=color)134 135    plt.tight_layout()136 137    # Textual summary138    text_summary = (139        f"**Dataset used**: `{final_id}`\n\n"140        f"**Label column**: `{label_column}`\n\n"141        f"**Feature columns**: `{feature_columns}`\n\n"142        f"**Accuracy**: {accuracy:.3f}\n\n"143    )144 145    return text_summary, fig146 147###############################################################################148# Gradio UI149###############################################################################150with gr.Blocks() as demo:151 152    # High-level title and description153    gr.Markdown(154        """155        # Introduction to Gradient Boosting156 157        This Space demonstrates how to train a [GradientBoostingClassifier](https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.GradientBoostingClassifier.html#gradientboostingclassifier) from **scikit-learn** on **tabular datasets** hosted on the [Hugging Face Hub](https://huggingface.co/datasets).158 159        Gradient Boosting is an ensemble machine learning technique that combines many weak learners (usually small decision trees) in an iterative, stage-wise fashion to create a stronger overall model. 160        In each step, the algorithm fits a new weak learner to the current errors of the combined ensemble, effectively allowing the model to focus on the hardest-to-predict data points. 161        By repeatedly adding these specialized trees, Gradient Boosting can capture complex patterns and deliver high predictive accuracy, especially on tabular data.162 163        **Put simply, Gradient Boosting makes a big deal out of small anomolies!**164 165        **Purpose**:166        - Easily explore hyperparameters (_learning_rate, n_estimators, max_depth_) and quickly train an ML model on real data.167        - Visualise model performance via confusion matrix heatmap and a feature importance plot.168 169        **Notes**:170        - The dataset must have a **"train"** split with tabular columns (i.e., no nested structures).171        - Large datasets may take time to download/train.172        - The confusion matrix helps you see how predictions compare to ground-truth labels. The diagonal cells show correct predictions; off-diagonal cells indicate misclassifications.173        - The feature importance plot shows which features the model relies on the most for its predictions.174 175        ---176 177        **Usage**:178        1. Select one of the suggested datasets from the dropdown _or_ enter any valid dataset from the [Hugging Face Hub](https://huggingface.co/datasets).179        2. Click **Load Columns** to retrieve the column names from the dataset's **train** split.180        3. Choose exactly _one_ **Label column** (the target) and one or more **Feature columns** (the inputs).181        4. Adjust hyperparameters (learning_rate, n_estimators, max_depth, test_size).182        5. Click **Train & Evaluate** to train a Gradient Boosting model and see its accuracy, feature importances, and confusion matrix.183 184        You are now a machine learning engineer, congratulations 🤗 185        186        ---187        """188    )189 190    with gr.Row():191        dataset_dropdown = gr.Dropdown(192            label="Choose suggested dataset",193            choices=SUGGESTED_DATASETS,194            value=SUGGESTED_DATASETS[0]195        )196        custom_dataset_id = gr.Textbox(197            label="Or enter a custom dataset ID",198            placeholder="e.g. user/my_custom_dataset"199        )200 201    load_cols_btn = gr.Button("Load Columns")202    load_cols_info = gr.Markdown()203 204    with gr.Row():205        label_col = gr.Dropdown(choices=[], label="Label column (choose 1)")206        feature_cols = gr.CheckboxGroup(choices=[], label="Feature columns (choose 1 or more)")207 208    # Model Hyperparameters209    learning_rate_slider = gr.Slider(210        minimum=0.01, maximum=1.0, value=0.1, step=0.01, 211        label="learning_rate"212    )213    n_estimators_slider = gr.Slider(214        minimum=50, maximum=300, value=100, step=50, 215        label="n_estimators"216    )217    max_depth_slider = gr.Slider(218        minimum=1, maximum=10, value=3, step=1, 219        label="max_depth"220    )221    test_size_slider = gr.Slider(222        minimum=0.1, maximum=0.9, value=0.3, step=0.1, 223        label="test_size fraction (0.1-0.9)"224    )225 226    train_button = gr.Button("Train & Evaluate")227 228    output_text = gr.Markdown()229    output_plot = gr.Plot()230 231    # Link the "Load Columns" button -> update_columns function232    load_cols_btn.click(233        fn=update_columns,234        inputs=[dataset_dropdown, custom_dataset_id],235        outputs=[label_col, feature_cols, load_cols_info],236    )237 238    # Link "Train & Evaluate" -> train_model function239    train_button.click(240        fn=train_model,241        inputs=[242            dataset_dropdown,243            custom_dataset_id,244            label_col,245            feature_cols,246            learning_rate_slider,247            n_estimators_slider,248            max_depth_slider,249            test_size_slider250        ],251        outputs=[output_text, output_plot],252    )253 254demo.launch()255