sentiment_price_finalFunction.py

def sentimentAndPrice(ticker, start, end, numtweets=20):
    """
    Visually compares sentiment with the closing price of a given stock ticker.
    """
    # Creating a DF that contains daily tweets between two dates
    df = tweetByDay(start, end, pd.DataFrame(), search="$"+ticker, limit=numtweets)
        
    # Analyzing the sentiment of each tweet
    sent_df = getSentiment(
        df, 
        measurement='compound'
    )
    
    # Getting stock price history
    stock_df = getStockPrices(
        ticker, 
        start, 
        end
    )
    
    # Merging the two DF
    comb_df = sent_df.merge(stock_df, how='outer', sort=True)
    
    # Shifting the sentiment scores 1 day to compensate for lookahead bias
    comb_df['sentiment'] = comb_df['sentiment'].shift(1)
        
    # How often sentiment matched return

    # Dropping NAs so they are not compared
    drop_df = comb_df.dropna()

    # Comparing matches
    match = (drop_df['sentiment'].apply(lambda x: x>0)==drop_df['returns'].apply(lambda x: x>0))

    # Counting instances where they match
    match = match.value_counts().rename({False: "Didn't predict return",
                                         True: "Successfully predicted return"}).to_frame()
    
    # Visualizing matches in sentiment and return
    fig = px.bar(
        match,
        x=0,
        y=match.index,
        color=match.index,
        title="Instances when Sentiment predicts Return",
        labels={"index": "Prediction",
                "0": "Count"}
    )

    fig.show()
    
        
    # Visualizing the sentiment and price
    fig = px.bar(
        comb_df,
        x='date',
        y=['returns', 'sentiment'],
        barmode='group',
        title=f"Returns & Sentiment over Time for {ticker}"
    )
    
    return fig.show()