Multivariate Time Series Analysis of Sentiment Scores

Gathering data from Twitter and NewsApi

Installs for Twint

pip install nest_asyncio

pip install --upgrade git+https://github.com/twintproject/twint.git@origin/master#egg=twint

Imports

import twint import nest_asyncio nest_asyncio.apply() c = twint.Config()

from newsapi import NewsApiClient

Set API and Client

api_key = "0ef1c61926f54984abcca4338225fd66" newsapi = NewsApiClient(api_key=api_key

Pull data from NewsAPI

pltr_headlines = newsapi.get_everything( q="PLTR", language="en", page_size=10, sort_by="relevancy", from_param="2021-02-29" )

pltr_newsapi_df = pd.DataFrame.from_dict(pltr_headlines["articles"]) pltr_newsapi_df.head()

Pull Twitter data using Twint and save to CSV file

c.Search= "$DOT" or "DOT.X" c.Since= "2020-09-29" c.Until = '2021-04-04' c.Limit= 1000 c.Lang= "en" c.Store_csv= True c.Output= "Search.csv"

twint.run.Search(c)

Clean data

df = pd.read_csv('Search.csv', encoding="utf-8-sig") df= df[["id", "created_at", "tweet", "language"]] df=df.loc[df["language"]=="en"] df=df.rename(columns={"id": "ID", "created_at": "Date", "tweet": "Tweet"}) df=df.drop(["language"], axis=1) df=df.set_index("Date") df.shape

Import local csv to colab

from google.colab import files uploaded = files.upload()

Create function to calculate sentiment based on compound score

def get_sentiment(score): """ Calculates the sentiment based on the compound score. """ result = 0 # Neutral by default if score >= 0.05: # Positive result = 1 elif score <= -0.05: # Negative result = -1 return result

Create empty list for Twitter sentiments

pltr_twtr_sentiments = []

Iterate through rows to analyze tweets

for index, row in pltr_twitter_df.iterrows(): try: text = row["Tweet"] date = row["Date"] sentiment = analyzer.polarity_scores(text) compound = sentiment["compound"] pos = sentiment["pos"] neu = sentiment["neu"] neg = sentiment["neg"] pol= get_sentiment(compound)

    # Append data to list
    pltr_twtr_sentiments.append({
        "text": text,
        "date": date,
        "compound": compound,
        "positive": pos,
        "negative": neg,
        "neutral": neu,
        "Polarity Score":pol
        
    })
    
except AttributeError:
    pass

import pltr_twitter_sentiment_df csv to colab

pltr_twitter_sentiment_df = pd.DataFrame(pltr_twtr_sentiments) pltr_twitter_sentiment_df.info()

lfarlima/Multivariate-Time-Series-Analysis-of-Sentiment-Scores