This project is based on Facebook/Meta available personal data for Messenger conversations. Here I collected, clean-up and plotted some statistics for a group chat with me and some friends. He had more than 400,000 messages over more than 5 years.
This simple analysis focus on the time statistics for each participant: how much message each participant sends for a given period of the day, which days of the week they are more active and how that changed with years.
I also considered the 'react' dynamics (emojis that one can react to messagens among ourselves): who reacted more, who gets more reacts and how the participants reacted among themselves.
Python
os, pandas, numpy, matplotlib, json and datetime
Facebook/Meta conversation as JSON files downloaded from here
Can be found in the folder results in this reposity. Plots generated using the notebook also in this repository.
- Fix the emoji encoding to plot the results using matplotlib
-
We are more active after 10am with message peaking around 8-9pm
-
Our messages are consistent for all days of week with more chat during 2020 (pandemic)
-
The overall reaction probability is around 4-12%
-
The most used emoji for react, by far, is 😆