app.py

import ccxt
import ccxt.async_support as ccxt
import pandas as pd
from datetime import datetime, timedelta
from prophet import Prophet
import matplotlib.pyplot as plt
import streamlit as st

# Initialize the exchange
binance = ccxt.bitget()
symbol = "BTC/USDT"

def fetch_btc_usdt_data(start_date, end_date):
    since = int(datetime.strptime(start_date, '%Y-%m-%d').timestamp() * 1000)
    end = int(datetime.strptime(end_date, '%Y-%m-%d').timestamp() * 1000)
    timeframe = '1d'  # Daily data
    data = []

    while since < end:
        ohlcv = binance.fetch_ohlcv(symbol, timeframe, since, limit=1000)
        if not ohlcv:
            break
        since = ohlcv[-1][0] + 1
        data.extend(ohlcv)

    # Convert data to DataFrame
    columns = ['timestamps', 'open', 'high', 'low', 'close', 'volume']
    df = pd.DataFrame(data, columns=columns)
    df['timestamps'] = pd.to_datetime(df['timestamps'], unit='ms')
    df.set_index('timestamps', inplace=True)

    # Return the DataFrame
    return df

def train_and_forecast(df):
    # Prepare the data for Prophet
    df_prophet = df[['close']].reset_index()
    df_prophet.rename(columns={'timestamps': 'ds', 'close': 'y'}, inplace=True)

    # Train the Prophet model
    model = Prophet(daily_seasonality=True)
    model.fit(df_prophet)

    # Make a future dataframe for predictions (e.g., next 30 days)
    future = model.make_future_dataframe(periods=30, freq='D')

    # Get the forecast
    forecast = model.predict(future)

    return forecast, model

def plot_forecast(forecast, model):
    # Plot the forecast
    fig, ax = plt.subplots(figsize=(10, 6))
    model.plot(forecast, ax=ax)
    
    # Plot the components
    fig2, ax2 = plt.subplots(figsize=(10, 6))
    model.plot_components(forecast)

    return fig, fig2

# Streamlit UI
st.title("BTC/USDT Price Forecasting")
st.markdown("""
This app uses Facebook's Prophet model to forecast the future prices of BTC/USDT based on historical data from Bitget. 
You can select the start and end dates to get a prediction for the next 30 days.
""")

# Date input
start_date = st.date_input('Start Date', datetime.today() - timedelta(days=365))
end_date = st.date_input('End Date', datetime.today())

if start_date and end_date:
    # Fetch the data
    df = fetch_btc_usdt_data(start_date.strftime('%Y-%m-%d'), end_date.strftime('%Y-%m-%d'))
    
    # Train and forecast
    forecast, model = train_and_forecast(df)
    
    # Plot the forecast and components
    st.subheader("Forecast Plot")
    fig1, fig2 = plot_forecast(forecast, model)
    
    st.pyplot(fig1)  # Forecast plot
    st.pyplot(fig2)  # Components plot