app.py

import gradio as gr
import pandas as pd
import numpy as np
import pickle

# Load models
model = pickle.load(open("logistic_model.pkl", "rb"))
scaler = pickle.load(open("scaler_lr.pkl", "rb"))
kmeans = pickle.load(open("kmeans_model.pkl", "rb"))
feature_columns = pickle.load(open("feature_columns.pkl", "rb"))

# Prediction function
def predict(price, rating, discount, category):

    input_df = pd.DataFrame(columns=feature_columns)
    input_df.loc[0] = 0

    input_df["discount_price"] = price
    input_df["ratings"] = rating
    input_df["discount_percent"] = discount

    cluster = kmeans.predict([[price, rating, 3, discount]])[0]

    cluster_col = f"cluster_{cluster}"

    if cluster_col in input_df.columns:
        input_df[cluster_col] = 1

    category_col = f"main_category_{category}"

    if category_col in input_df.columns:
        input_df[category_col] = 1

    input_scaled = scaler.transform(input_df)

    prob = model.predict_proba(input_scaled)[0][1]

    prob_percent = round(prob * 100, 2)

    if prob > 0.7:
        recommendation = "Strong demand expected. Pricing is optimal."
    elif prob > 0.4:
        recommendation = "Moderate demand. Consider adjusting price or discount."
    else:
        recommendation = "Low demand probability. Consider lowering price."

    return f"{prob_percent}%", recommendation


# Interface
interface = gr.Interface(

    fn=predict,

    inputs=[
        gr.Number(label="Product Price (₹)", value=1000),
        gr.Slider(0,5,value=4,label="Rating"),
        gr.Slider(0,90,value=30,label="Discount %"),
        gr.Dropdown(
            [
                "beauty & health",
                "grocery & gourmet foods",
                "home & kitchen",
                "kids' fashion",
                "men's shoes",
                "stores",
                "toys & baby products",
                "tv, audio & cameras"
            ],
            label="Category"
        )
    ],

    outputs=[
        gr.Text(label="High Demand Probability"),
        gr.Text(label="Pricing Recommendation")
    ],

    title="Amazon Pricing Optimization System",

    description="Predict product demand probability based on pricing, rating, discount, and category."

)

interface.launch()