IPython notebook of a behavioural biometrics analysis on key logger data.
This notebook was created to explore the relationship (if any!) between user emotions and/or fatigue level and keyboard and mouse usage.
I use Pandas for data analysis, and the Seaborn package for data visualization. The keyboard and mouse capture data was originally kept in SQLite and I have stuck with that.
Here is the simple experiment layout. The typing of these instructions was actually the data to be captured:
Start keylogger.
Type up this experimental outline.
Calm User, just typing something up. A few instructions, a message maybe. Click to enlarge text window.
A bit of window movement. Dragging text window around the page.
Bored User: tracks mouse back and forth …
User reading document with mouse to help.
User highlights data to read it.
User loses mouse between screens.
User with extremem mouse jerk.
User picking up mouse to move its postion on the screen.
User angrily smashes key board.
Jk,uiolhjmASopwe-1jk,zsdjik,lxsfdfgbrfvd78uikj kl;'/tgyh xfdc
Some more mouse jerk!
Typing a sfast as I can because I am hurridly trying to get a message out to someone, seriously as fast as I can > and somehow very accurately as well.
Perhaps a user who clicks a lot when bored or annoyed. Or they just can't wait when pushing a "button".
Turn of keylogger.