app.py

# Import necessary libraries
import pandas as pd
import matplotlib.pyplot as plt
from statsmodels.tsa.arima.model import ARIMA
from sklearn.preprocessing import StandardScaler
import streamlit as st
import pickle
import numpy as np
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.metrics.pairwise import cosine_similarity
import nltk
from nltk.corpus import stopwords
from nltk.tokenize import word_tokenize
import os  # Added for file path handling
import openai


openai.api_key = "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"

# Download the necessary NLTK datasets
nltk.download('punkt')
nltk.download('stopwords')

# Function to preprocess data
def preprocess_data(df):
    # Handle missing values by forward filling
    df = df.fillna(method='ffill')

    # Optionally, handle any remaining missing values by backward filling
    df = df.fillna(method='bfill')

    # Optionally, standardize (scale) the data to have zero mean and unit variance
    scaler = StandardScaler()
    df[df.columns] = scaler.fit_transform(df[df.columns])

    return df

def trading_advice(actual, prediction):
    """
    Provide trading advice based on prediction and actual prices.
    """
    if prediction > actual:
        return "Based on the predictions, you might consider buying."
    elif prediction < actual:
        return "Based on the predictions, you might consider selling."
    else:
        return "The price seems stable. You might consider holding."

def chatbot_response(user_input, predefined_responses):
    # Tokenize and vectorize user input and predefined responses
    vectorizer = TfidfVectorizer(tokenizer=lambda text: nltk.word_tokenize(text, language='english'),
                                 stop_words=stopwords.words('english'))
    vectors = vectorizer.fit_transform([user_input] + predefined_responses)

    # Calculate cosine similarities
    cosine_matrix = cosine_similarity(vectors)

    # Find the most similar predefined response to the user's input
    response_idx = np.argmax(cosine_matrix[0][1:])

    return predefined_responses[response_idx]

def get_gpt3_response(user_input):
    prompt = f"Answer questions related to cryptocurrency.\n\nUser: {user_input}\nBot:"
    response = openai.Completion.create(
        engine="text-davinci-002",
        prompt=prompt,
        max_tokens=50  # You can adjust this based on response length
    )
    return response.choices[0].text


# Streamlit app
def main():
    st.title("Cryptocurrency Price Prediction and Trading Bot")

    # Specify the path to the CSV file
    csv_path = './crypto_dataset (9).csv'  # Update with your specific file path

    # Check if the CSV file exists at the specified path
    if not os.path.exists(csv_path):
        st.error("The specified CSV file does not exist. Please provide a valid file path.")
        return

    # Read and preprocess the input data
    input_data = pd.read_csv(csv_path, parse_dates=['Timestamp'], index_col='Timestamp')
    input_data = preprocess_data(input_data)

    # List of coins
    coins = ['BTC-USD Close', 'ETH-USD Close', 'LTC-USD Close']

    # Get user's choice of cryptocurrency
    coin_choice = st.selectbox("Select a cryptocurrency", coins)

    # Train ARIMA model if needed
    if coin_choice == "BTC-USD Close":
        coin_model_path = 'btc-usd_close_model.pkl'  # Update with your desired model path
        coin_column = 'BTC-USD Close'
    elif coin_choice == "ETH-USD Close":
        coin_model_path = 'eth-usd_close_model.pkl'  # Update with your desired model path
        coin_column = 'ETH-USD Close'
    else:
        coin_model_path = 'ltc-usd_close_model.pkl'  # Update with your desired model path
        coin_column = 'LTC-USD Close'

    # Check if the model exists, if not, train it
    if not os.path.exists(coin_model_path):
        # Split the data into training and testing sets
        train_size = int(len(input_data) * 0.8)
        train, test = input_data[:train_size], input_data[train_size:]

        # Build and train the ARIMA model
        model = ARIMA(train[coin_column], order=(5, 1, 0))  # Example order, you can tune this
        model_fit = model.fit()

        # Save the trained model using pickle
        with open(coin_model_path, 'wb') as model_file:
            pickle.dump(model_fit, model_file)

    # Load the model using pickle
    with open(coin_model_path, 'rb') as model_file:
        model_fit = pickle.load(model_file)

    # Make predictions
    predictions = model_fit.forecast(steps=len(input_data))

    # Chatbot live interaction

    st.header("Chat with Trading Bot")

    user_message = st.text_input("You: ")
    predefined_responses = [
        "The predicted price for the next period is {}.".format(predictions.iloc[-1]),
        trading_advice(input_data[coin_column].iloc[-1], predictions.iloc[-1]),
        "I am here to help with your cryptocurrency trading decisions.",
        "Can you specify your query?"
    ]

    if user_message:
        if "predict" in user_message.lower() or "forecast" in user_message.lower():
            bot_reply = trading_advice(input_data[coin_column].iloc[-1], predictions.iloc[-1])
        else:
            bot_reply = get_gpt3_response(user_message)
        st.write(f"Bot: {bot_reply}")
        
    # Display predictions
    st.header("Price Predictions")
    st.write(predictions)

    # Visualize results
    st.header("Price Prediction Chart")
    fig, ax = plt.subplots(figsize=(10, 6))
    ax.plot(input_data.index, input_data[coin_column], label='Actual')
    ax.plot(input_data.index, predictions, label='Predicted', linestyle='--')
    ax.set_title(f'{coin_choice} Price Prediction')
    ax.legend()
    st.pyplot(fig)

    

if __name__ == "__main__":
    main()