Knight-coderr/StockAnalysis
12
1import streamlit as st2import pandas as pd3import yfinance as yf4from textblob import TextBlob5import joblib6import matplotlib.pyplot as plt7from datetime import datetime8import plotly.express as px9 10# Function to load stock data using yfinance/ this is going to refresh after 1 day11@st.cache_data(ttl=86400)12def load_stock_data(tickers, start_date, end_date):13 with st.spinner('Downloading stock data...'):14 data = yf.download(tickers, start=start_date, end=end_date, group_by='ticker', auto_adjust=True)15 16 all_data = []17 for ticker in tickers:18 df = data[ticker].copy().reset_index()19 df['Stock Name'] = ticker20 all_data.append(df)21 22 merged_data = pd.concat(all_data, ignore_index=True)23 return merged_data24 25tickers = ['TSLA', 'MSFT', 'PG', 'META', 'AMZN', 'GOOG', 'AMD', 'AAPL', 'NFLX', 'TSM',26 'KO', 'F', 'COST', 'DIS', 'VZ', 'CRM', 'INTC', 'BA', 'BX', 'NOC', 'PYPL', 'ENPH', 'NIO', 'ZS', 'XPEV']27start_date = (datetime.today() - pd.DateOffset(years=1)).strftime('%Y-%m-%d')28end_date = datetime.today().strftime('%Y-%m-%d')29 30# Cache stock data for 1 day using st.cache_data31@st.cache_data(ttl=86400)32def load_and_cache_stock_data():33 return load_stock_data(tickers, start_date, end_date)34 35# Initialize stock_data once at app startup36if "stock_data" not in st.session_state:37 st.session_state["stock_data"] = load_and_cache_stock_data()38 39stock_data = st.session_state["stock_data"]40 41 42# Perform sentiment analysis on tweets (assuming you still have your tweets data)43tweets_data = pd.read_csv('data/stock_tweets.csv')44 45# Convert the Date columns to datetime46tweets_data['Date'] = pd.to_datetime(tweets_data['Date']).dt.date47 48# Perform sentiment analysis on tweets49def get_sentiment(tweet):50 analysis = TextBlob(tweet)51 return analysis.sentiment.polarity52 53tweets_data['Sentiment'] = tweets_data['Tweet'].apply(get_sentiment)54 55# Aggregate sentiment by date and stock56daily_sentiment = tweets_data.groupby(['Date', 'Stock Name']).mean(numeric_only=True).reset_index()57 58# Convert the Date column in daily_sentiment to datetime64[ns]59daily_sentiment['Date'] = pd.to_datetime(daily_sentiment['Date'])60 61# Merge stock data with sentiment data62merged_data = pd.merge(stock_data, daily_sentiment, how='left', on=['Date', 'Stock Name'])63 64# Fill missing sentiment values with 0 (neutral sentiment)65merged_data['Sentiment'] = merged_data['Sentiment'].fillna(0)66 67# Sort the data by date68merged_data = merged_data.sort_values(by='Date')69 70# Create lagged features71merged_data['Prev_Close'] = merged_data.groupby('Stock Name')['Close'].shift(1)72merged_data['Prev_Sentiment'] = merged_data.groupby('Stock Name')['Sentiment'].shift(1)73 74# Create moving averages75merged_data['MA7'] = merged_data.groupby('Stock Name')['Close'].transform(lambda x: x.rolling(window=7).mean())76merged_data['MA14'] = merged_data.groupby('Stock Name')['Close'].transform(lambda x: x.rolling(window=14).mean())77 78# Create daily price changes79merged_data['Daily_Change'] = merged_data['Close'] - merged_data['Prev_Close']80 81# Create volatility82merged_data['Volatility'] = merged_data.groupby('Stock Name')['Close'].transform(lambda x: x.rolling(window=7).std())83 84# Drop rows with missing values85merged_data.dropna(inplace=True)86 87# Load the best model88model_filename = 'model/best_model.pkl'89model = joblib.load(model_filename)90 91# Streamlit application layout92st.title("Stock Price Prediction Using Sentiment Analysis")93 94# User input for stock data95st.header("Input Stock Data")96stock_names = merged_data['Stock Name'].unique()97selected_stock = st.selectbox("Select Stock Name", stock_names)98days_to_predict = st.number_input("Number of Days to Predict", min_value=1, max_value=30, value=10)99 100# Get the latest data for the selected stock101latest_data = merged_data[merged_data['Stock Name'] == selected_stock].iloc[-1]102prev_close = latest_data['Close']103prev_sentiment = latest_data['Sentiment']104ma7 = latest_data['MA7']105ma14 = latest_data['MA14']106daily_change = latest_data['Daily_Change']107volatility = latest_data['Volatility']108 109# Display the latest stock data in a table110latest_data_df = pd.DataFrame({111 'Metric': ['Previous Close Price', 'Previous Sentiment', '7-day Moving Average', '14-day Moving Average', 'Daily Change', 'Volatility'],112 'Value': [prev_close, prev_sentiment, ma7, ma14, daily_change, volatility]113})114 115st.write("Latest Stock Data:")116st.write(latest_data_df)117 118st.write("Use the inputs above to predict the next days close prices of the stock.")119if st.button("Predict"):120 predictions = []121 latest_date = datetime.now()122 123 for i in range(days_to_predict):124 X_future = pd.DataFrame({125 'Prev_Close': [prev_close],126 'Prev_Sentiment': [prev_sentiment],127 'MA7': [ma7],128 'MA14': [ma14],129 'Daily_Change': [daily_change],130 'Volatility': [volatility]131 })132 133 next_day_prediction = model.predict(X_future)[0]134 predictions.append(next_day_prediction)135 136 # Update features for next prediction137 prev_close = next_day_prediction138 ma7 = (ma7 * 6 + next_day_prediction) / 7 # Simplified rolling calculation139 ma14 = (ma14 * 13 + next_day_prediction) / 14 # Simplified rolling calculation140 daily_change = next_day_prediction - prev_close141 142 # Prepare prediction data for display143 prediction_dates = pd.date_range(start=latest_date + pd.Timedelta(days=1), periods=days_to_predict)144 prediction_df = pd.DataFrame({145 'Date': prediction_dates,146 'Predicted Close Price': predictions147 })148 149 st.subheader("Predicted Prices")150 st.dataframe(prediction_df)151 152 # Plot predictions using Plotly153 fig = px.line(prediction_df, x='Date', y='Predicted Close Price', markers=True, title=f"{selected_stock} Predicted Close Prices")154 st.plotly_chart(fig, use_container_width=True)155 156 # ----------------------------------------157 # Enhanced Visualizations158 st.header("Enhanced Stock Analysis")159 stock_history = merged_data[merged_data['Stock Name'] == selected_stock]160 161 # Date filter slider162 min_date = pd.to_datetime(merged_data['Date'].min()).date()163 max_date = pd.to_datetime(merged_data['Date'].max()).date()164 165 date_range = st.slider(166 "Select Date Range for Visualizations",167 min_value=min_date,168 max_value=max_date,169 value=(min_date, max_date),170 format="YYYY-MM-DD"171 )172 173 filtered_data = stock_history[(stock_history['Date'] >= pd.to_datetime(date_range[0])) & 174 (stock_history['Date'] <= pd.to_datetime(date_range[1]))]175 176 with st.expander("Price vs Sentiment Trend"):177 fig1 = px.line(filtered_data, x='Date', y=['Close', 'Sentiment'],178 labels={'value': 'Price / Sentiment', 'variable': 'Metric'},179 title=f"{selected_stock} - Close Price & Sentiment")180 st.plotly_chart(fig1, use_container_width=True)181 182 with st.expander("Volatility Over Time"):183 fig2 = px.line(filtered_data, x='Date', y='Volatility',184 title=f"{selected_stock} - 7-Day Rolling Volatility")185 st.plotly_chart(fig2, use_container_width=True)186 187 with st.expander("Moving Averages (MA7 vs MA14)"):188 fig3 = px.line(filtered_data, x='Date', y=['MA7', 'MA14'],189 labels={'value': 'Price', 'variable': 'Moving Average'},190 title=f"{selected_stock} - Moving Averages")191 st.plotly_chart(fig3, use_container_width=True)192 193 with st.expander("Daily Price Change"):194 fig4 = px.line(filtered_data, x='Date', y='Daily_Change',195 title=f"{selected_stock} - Daily Price Change")196 st.plotly_chart(fig4, use_container_width=True)197 198 with st.expander("Sentiment Distribution"):199 fig5 = px.histogram(filtered_data, x='Sentiment', nbins=30,200 title=f"{selected_stock} - Sentiment Score Distribution")201 st.plotly_chart(fig5, use_container_width=True)202