saiki0007 / crypto_sentiment_analysis

Summary

Data:

• Data(messages) was obtained from a crypto group on telegram between the dates 05/01/2021 and 05/15/2021 (inclusive)

Preprocessing:

• First, the message data is cleaned to remove hyperlinks, brackets and digits from messages.
• Then the data was analyzed using the library 'langdetect' and messages in english language were extracted.
• Then only the messages containing "SHIB" or "DOGE" were extracted.

Sentiment Analysis:

• TextBlob library was used to perform the sentiment analysis.
• This library function outputs a sentiment score for each message, where '>0' is considered as positive, '0' is considered neutral and '<0' is considered negative emotion.

Results:

• The total number of messages, positive, negative, and neutral sentiments, and average sentiment score per day are extracted into a dataframe.
• Plotly is used to plot average sentiment score vs Date and Number of messages per day vs Date
• 
• 

Running the code

• Install the packages in requirements.txt by running:

pip install -r requirements.txt

• Run the python file "crypto_sentiment_analysis" for results and plots:

python3 crypto_sentiment_analysis.py

• Note: Make sure to include "messages.json" in the same directory