app.py

# app.py
import re, string
import joblib
import numpy as np
import gradio as gr

# -----------------------------
# 1️⃣ Load model and vectorizer
# -----------------------------
model = joblib.load("svm_model.pkl")
tfidf = joblib.load("tfidf_vectorizer.pkl")

# -----------------------------
# 2️⃣ Preprocess function
# -----------------------------
def preprocess(text):
    text = str(text).lower()
    text = re.sub(f"[{string.punctuation}]", "", text)
    text = re.sub(r"\d+", "", text)
    text = text.strip()
    return text

# -----------------------------
# 3️⃣ Prediction function
# -----------------------------
def predict_sentiment(review):
    if not review:
        return "Error: No review provided", 0.0
    cleaned = preprocess(review)
    vectorized = tfidf.transform([cleaned])
    
    pred = model.predict(vectorized)[0]
    sentiment = "positive" if pred == 1 else "negative"
    
    confidence = model.decision_function(vectorized)[0]
    confidence = 1 / (1 + np.exp(-confidence))  # sigmoid for probability-like score
    
    return sentiment, round(float(confidence)*100, 2)

# -----------------------------
# 4️⃣ Gradio UI
# -----------------------------
iface = gr.Interface(
    fn=predict_sentiment,
    inputs=gr.Textbox(label="Write a review here...", lines=5, placeholder="Type your review..."),
    outputs=[
        gr.Label(label="Predicted Sentiment"),
        gr.Number(label="Confidence (%)")
    ],
    title="Amazon Review Sentiment Analysis",
    description="Enter an Amazon product review and get the predicted sentiment along with confidence score.",
    theme="default"
)

# -----------------------------
# 5️⃣ Launch
# -----------------------------
if __name__ == "__main__":
    iface.launch()