This idea based on Zack's tweet
Let me repeat it here:
run last 500 tweets through shitty sentiment analysis and give a score between -1 (negative) and 1(positive)
I also want to generate a special pretty badge that you can post to Twitter that will display a special trait like 😇 or 😈
Sentiment analysis of the tweets dataset
Some threads collect thousands likes, but seems like no one really going through the links. For example this kaggle about text mining slowly going with string manipulation and regexes, but interesting classification at the end fails because some names are not defined 😖
Nevertheless it uses very interesting dataset. Guys from CrowdFlower decided to gues gender by text we write. Amazing, right?
CrowdFlower team: to run a text mining experiment we pulled 20,000 tweets with the word “the” and the word “and”.
me: Why?
CrowdFlower team:
Reading about it here: https://kdnuggets.com/2015/11/machine-learning-predict-gender.html
Data on kaggle: https://kaggle.com/datasets/crowdflower/twitter-user-gender-classification