app.py

import gradio as gr
import pandas as pd
import numpy as np
import yfinance as yf
from prophet import Prophet
from prophet.plot import plot_plotly, plot_components_plotly
import plotly.graph_objects as go
#from sklearn.metrics import mean_absolute_error, mean_squared_error


import warnings
warnings.simplefilter(action='ignore', category=FutureWarning)


stocks = pd.read_excel('Stocks.xlsx', usecols =[1,2,3])

period_options = {
    "1wk": "1 Week",
    "1mo": "1 Month",
    "1y": "1 Year",
    "5y": "5 Years"
}


# Create a Gradio radio button group for the period
period = gr.Radio(label="Training Period: ", choices=list(period_options.values()), value="1 Week")



def get_forecast(company_name):
    symbol_nse = stocks[stocks['Company Name'] == company_name]['Symbol'].values[0] + '.NS'
    #period_key = [key for key, value in period_options.items() if value == period][0]
    #stock_df = yf.download(symbol_nse, period = period_key)
    
    stock_df = yf.download(symbol_nse, period = '5y')
    stock_df.drop(stock_df.columns[[0,1,2,4,5]], axis=1, inplace=True)
    stock_df.reset_index(inplace=True)
    stock_df.columns = ['ds', 'y']
    #stock_df = stock_df[['ds', 'y', 'cap']]
    
    #model = Prophet(growth='logistic')
    model = Prophet()
    model.fit(stock_df)
    
    future = model.make_future_dataframe(periods = 7)
    forecast = model.predict(future)
    forecast_df = forecast[['ds', 'yhat', 'yhat_lower', 'yhat_upper']]
    
    # Calculate accuracy metrics
    actual_prices = stock_df['y'].values[-1000:]  # Last 1000 actual prices
    predicted_prices = forecast_df['yhat'].values[-1000:]  # Forecasted prices for the last 1000 days
    #mae = mean_absolute_error(actual_prices, predicted_prices)
    #mse = mean_squared_error(actual_prices, predicted_prices)
    #rmse = np.sqrt(mse)
    mape = np.mean(np.abs((actual_prices - predicted_prices) / actual_prices)) * 100
    accuracy = 100 - mape
    
    
    fig = plot_plotly(model, forecast_df, xlabel = "Date", ylabel = "Price", figsize=(1400,800))
    #fig.update_layout(autosize=True)
    #fig.update_xaxes(automargin =True)
    
    # Display accuracy metrics
    # accuracy_text = f"MAE: {mae:.2f}, MSE: {mse:.2f}, RMSE: {rmse:.2f}, MAPE: {mape:.2f}%"
    accuracy_text = f"{accuracy:.2f} % based on past 03 Years Prediction vs Actual Stock Price."  
       
    return fig , accuracy_text


with gr.Blocks() as demo:
    
    gr.Markdown(
        """
        # Stock Price Trend Prediction - using PROPHET Model
        Select the Stock from Dropdown Menu to get Next Week Prediction
        
        """
        
    )
    with gr.Row():
        dropdown = gr.Dropdown(label="Company Name", choices=stocks['Company Name'].tolist(), filterable = True, info = 'Select NSE Stock')
    
    with gr.Row():
        with gr.Column():
            accuracy_textbox = gr.Textbox(label="Model Accuracy", visible=True, info = "Accuracy above 95% is considerate", interactive = False)
        with gr.Column():
            None
        with gr.Column():
            None
        with gr.Column():    
            submit_btn = gr.Button(value = "Predict")
    
    
       
    gr.Markdown(
        """
        ### Select the Plot-Area to check Prediction for Next Week
        
        """
        
    )
    
          
    with gr.Row():
        forecast_plot = gr.Plot()
    
        
    
    submit_btn.click(get_forecast, inputs=dropdown, outputs=[forecast_plot, accuracy_textbox])            
    
      
demo.launch(share=True)