import gradio as gr
import numpy as np
from sklearn.datasets import load_iris
from sklearn.ensemble import GradientBoostingClassifier
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score, confusion_matrix

# 1. Load dataset
iris = load_iris()
X, y = iris.data, iris.target
feature_names = iris.feature_names
class_names = iris.target_names

# Split into train/test
X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.3, random_state=42
)

# 2. Define a function that trains & evaluates a model given hyperparameters
def train_and_evaluate(learning_rate, n_estimators, max_depth):
    clf = GradientBoostingClassifier(
        learning_rate=learning_rate,
        n_estimators=n_estimators,
        max_depth=int(max_depth),
        random_state=42
    )
    clf.fit(X_train, y_train)
    y_pred = clf.predict(X_test)

    accuracy = accuracy_score(y_test, y_pred)
    cm = confusion_matrix(y_test, y_pred)
    cm_display = "\n".join([str(row) for row in cm])

    return f"Accuracy: {accuracy:.3f}\nConfusion Matrix:\n{cm_display}"

# 3. Define a prediction function for user-supplied feature values
def predict_species(sepal_length, sepal_width, petal_length, petal_width,
                    learning_rate, n_estimators, max_depth):
    clf = GradientBoostingClassifier(
        learning_rate=learning_rate,
        n_estimators=n_estimators,
        max_depth=int(max_depth),
        random_state=42
    )
    clf.fit(X_train, y_train)

    user_sample = np.array([[sepal_length, sepal_width, petal_length, petal_width]])
    prediction = clf.predict(user_sample)[0]
    return f"Predicted species: {class_names[prediction]}"

# 4. Build the Gradio interface
with gr.Blocks() as demo:
    with gr.Tab("Train & Evaluate"):
        gr.Markdown("## Train a GradientBoostingClassifier on the Iris dataset")
        learning_rate_slider = gr.Slider(0.01, 1.0, value=0.1, step=0.01, label="learning_rate")
        n_estimators_slider = gr.Slider(50, 300, value=100, step=50, label="n_estimators")
        max_depth_slider = gr.Slider(1, 10, value=3, step=1, label="max_depth")

        train_button = gr.Button("Train & Evaluate")
        output_text = gr.Textbox(label="Results")

        train_button.click(
            fn=train_and_evaluate,
            inputs=[learning_rate_slider, n_estimators_slider, max_depth_slider],
            outputs=output_text,
        )

    with gr.Tab("Predict"):
        gr.Markdown("## Predict Iris Species with GradientBoostingClassifier")
        sepal_length_input = gr.Number(value=5.1, label=feature_names[0])
        sepal_width_input = gr.Number(value=3.5, label=feature_names[1])
        petal_length_input = gr.Number(value=1.4, label=feature_names[2])
        petal_width_input = gr.Number(value=0.2, label=feature_names[3])
        
        # Hyperparams for the model that will do the prediction
        learning_rate_slider2 = gr.Slider(0.01, 1.0, value=0.1, step=0.01, label="learning_rate")
        n_estimators_slider2 = gr.Slider(50, 300, value=100, step=50, label="n_estimators")
        max_depth_slider2 = gr.Slider(1, 10, value=3, step=1, label="max_depth")

        predict_button = gr.Button("Predict")
        prediction_text = gr.Textbox(label="Prediction")

        predict_button.click(
            fn=predict_species,
            inputs=[
                sepal_length_input,
                sepal_width_input,
                petal_length_input,
                petal_width_input,
                learning_rate_slider2,
                n_estimators_slider2,
                max_depth_slider2,
            ],
            outputs=prediction_text
        )

demo.launch()