Predict Twitter Users Age

Recently I worked on an interesting prediction problem. It was predicting Twitter users age. I have been collecting twitter data for a research project. But just for fun I decided to play around with this data set. So I started to see if I can predict the demographics of the twitter user from tweets. I started with age.

At some point I will write a longer post realted to how I build the model, my error rates etc. But here is a plot of predicted ages where X axis showing ages and Y axis showing how many users per age. This prediction from model unseen data. Its interesting that data is skewed towards to common intuition where younger people are active in online communities.