R-User Group Meetup Event at SWISS

Schedule:
6.15 pm Door opens
6.30pm Welcome Zurich RUG
6.35pm Short introduction to Advanced Analytics Network at
SWISS
6.40pm Talk 1 by Benjamin Perucco (Abstracts: see below)
7.10pm Talk 2 by Joao Banha
7.40pm RUG Pitch Session: Are you hiring? Looking for an R related
job? ... or organising an interesting event? Take the stage
for one minute. (DM us if you need a slot!)
8.00pm Apéro sponsored by SWISS

***

How to get to the Venue:
Our office can be reached by bus 737 (direction Obstgartenstrasse) from the airport bus station or by a short walk from Kloten train station.
Detailed description (in German): https://tinyurl.com/y3grjrnz
Map of the SWISS headquarters: https://tinyurl.com/yy87j75m

Room:Charly Building, 2. Stock, Raum "PARIS" (C2-58)

***
Speaker 1: Benjamin Perucco | Data Analyst @ Zurich Airport

Title: Predicting the departing runway of a flight based on a classification tree model

Zurich airport’s capacity is restricted by highly complex runway and airspace dependencies. One of these dependencies exists between departing aircraft in southerly direction on runway 16 and arriving traffic from the north on runway 14. Because of this, and during times of high departure demand on runway 16, the arrival flow is regulated by the air navigation service provider. These regulations allow for departures on runway 16 to take off without congesting the airspace with arriving aircraft. However, for such a regulation to be effective, it must be set in place well in advance, in fact before aircraft actually have departed at their origin. Therefore, a reliable estimate on expected departure demand on runway 16 a couple of hours ahead highly supports the decision whether regulations are necessary or not.

We present several models to predict the departing runway for aircrafts in order to support the air navigation service provider with this decision. The latest of these models incorporates data from different sources like flight plan and weather model data and resulted in a classification tree model which can predict the runway with high accuracy. Besides insight on the steps undertaken for building the model, we also provide an outlook on how such models can help to improve operations at Zurich airport in the future.

--

Speaker 2: Joao Banha | Business Analyst @ Swiss International Air Lines

Title: NOTAM message classifier

A notice to airmen (NOTAM) is a semi-encoded message to alert personnel and systems concerned with flight operations of potential hazards along a flight route. Each year over 1 million NOTAM messages are issued worldwide and this number is growing. The messages contain crucial information for pilots, yet their number poses an increasing operational challenge, when messages have to be processed and classified manually.
We created a machine learning binary classifier in R to deal with this challenge. This classifier provides a recommendation on the overall importance of these messages. Exploring the data resulted in the conclusion that a large fraction of messages can be automatically classified with high confidence. R Shiny is also used for the visualization of results.