This aims to perform sentiment analysis on COVID-19 tweets using various classification models. We preprocess the data, convert words to vectors, and train models such as Naïve Bayes, SVM, and KNN. Finally, we compare their performance to determine the most accurate model for predicting sentiment in COVID-19 tweets.
Problem statement:
Create a classification model to predict the sentiment either (Positive or Negative) based on Covid Tweets
Context:
The tweets have been pulled from Twitter and manual tagging has been done then. The names and usernames have been given codes to avoid any privacy concerns.
The columns are described as follows:
- UserName: UserName in encrypted numbers
- ScreenName: ScreenName in encrypted numbers
- Location: Country from where tweet was pulled from
- TweetAt: Twee time
- OriginalTweet: Tweet content
- Sentiment: Positive, Negative, Neutral, Extremely Positive, Extremely Negative
Steps Involved:
- Read the dataset with encoding parameter set to ‘latin1’
- Remove handle null values (if any).
- Pre-Process the Covid tweets based on the following parameter: a) Tokenizing words b) Convert words to lower case c) Removing Punctuations d) Removing Stop words e) Stemming or lemmatizing the words
- Convert the 'Extremely Positive' and 'Extremely Negative' Sentiments to 'Positive' and 'Negative' sentiments respectively
- Transform the words into vectors using a) Count Vectorizer or b) TF-IDF Vectorizer
- Split data into training and test data.
- Apply the following models on the training dataset and generate the predicted value for the test dataset a) Multinomial Naïve Bayes Classification b) SVM Classification c) KNN Classification
- Predict the Sentiment for test data
- Compute Confusion matrix and classification report for each of these models.
- Report the model with the best accuracy.