Talking kukas and motikas in multivariate time-series

Multivariate time series are both pain and bread and butter of every single data scientist working for big shots, like banks. However, there we are - safety guys who want to do the same with FDR (flight data recorder, "black box") data. But, some aircraft record just eight parameters, the other ones thousands. Sometimes, rather always, we have to deal with missing individual data points. On the other hand, the signal can be missing for the duration of the flight and is back on afterwards. Then, the naming conventions: Airbus and Boeing - two completely different approaches to aircraft design. Still, not a word about the data analysis - but there will be.

About lecturer: Marko Vasiljevski is a physicist leading the data science department at Flight data services. His everyday work is about exploration of all available sources of aviation data, algorithm development, statistics, chatting with people and research. For data exploration and algorithm development he mainly uses open source tools like R, Python and PostgreSQL. The chats are mainly with fellow data scientists, pilots, developers and airlines’ flight safety officers, whilst the research covers reading the material on forgotten and novel approaches in machine learning, statistics and programming. Having fun is a very important aspect of his life and work, but he takes aviation safety extremely seriously and is keen to help airlines with interpretation and making use of their data. At flight safety and data science seminars and conferences, Marko is a speaker and promoter of safe flying, data sharing and data-driven decision making.