From Prototype to Operative Software with RapidMiner Data Analytics @ Lufthansa
Agenda 1 Lufthansa Industry Solutions: Who We Are 2 Our Daily Challenges 3 FlightPrediction @ Lufthansa 4 Lufthansa Industry Solutions & RapidMiner 2 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
IT Consulting and We are Systems Integration Lufthansa Lufthansa Industry Solutions combines the dynamics of an SME Industry with the economically secure Solutions background of Lufthansa; an internationally acting global corporation. 3 9/1/2017 Company Presentation
Services overview Lufthansa Industry Solutions Strategy and Management Consulting IT Service Provider Organization & IT System Operations of full Process Consulting Operations business processes IT Consulting IT System Strategy (processes, Integration & Consulting technologies, Process Development Organi- Application infrastructure) BPO Consulting / Infra- sational Mgmt. & Process structure specialized provider structure Operations Organization IT Consulting and System Integration 4 9/1/2017 Company Presentation
Wolfsburg N orderstedt Berlin Hamburg Wetzlar Our Basel Cologne Oldenburg Locations Miami Frankfurt 5 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
Hard Facts Wolfsburg N orderstedt Berlin 208 Mil. € Managing Director Hamburg Wetzlar Our Bernd Appel total revenue (46% of which within the Basel Lufthansa Group) Founding Cologne Oldenburg Locations 1998 as Industry Solutions >1300 division within Lufthansa Miami Frankfurt Systems AG trained employees 100 % owned by >200 Lufthansa customers 6 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
Passenger Aviation Finance and Service Digital Air Freight Airlines Industry Companies Service Companies Lufthansa Lufthansa Lufthansa Lufthansa Swiss Lufthansa LSG Sky Lufthansa Industry AirPlus LGBS Industry Cargo Passenger Austrian Technik Chefs Systems Solutions Solutions 7 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
Many Projects Successfully Delivered 8 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
About Me Quick Facts Past Present Future Predictive analytics: Probabilistic modelling Time series analysis Classification, Bayesian learning Learning theory regression, recommender systems, probabilistic modelling, … RapidMiner, R, Python, Fabian Werner Born 1985 PHP, Teradata, Oracle, Hadoop , … Ph.D. Mathematics (Number theory) 9 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
Our Daily Challenges Lufthansa Internal (Deep) Machine Dashboarding Predictive Analytics Data Ingestion Learning Airplane punctuality Arrival times Document and image Live flight data Competitor information Passenger connection Weather data streams recognition Airfreight load monitor Fuel & Weight forecast Damage recognition Airplane sensor data Predictive maintenance via audio data Recommend fligths Bots (see Mildred) 10 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
Process Model Scoping Phase Laboratory Implementation stage stage Laboratory stage Implementation stage 11 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
FlightPrediction @ Lufthansa We predict the arrival time on aircraft takeoff
The Cycle 13 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
Goal & Process Prediction Prediction Time Point Gate - to - Gate Rolling Airtime Rolling Departure Takeoff Landing Arrival 14 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
Goal & Process Prediction Prediction Time Point Gate - to - Gate Rolling Airtime Rolling Departure Takeoff Landing Arrival 15 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
Goal & Process Prediction Prediction Time Point Gate - to - Gate Rolling Airtime Rolling Departure Takeoff Landing Arrival Accuracy of existing prediction Avg. deviation estimated vs. actual OnBlock 16 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
Goal & Process Prediction Prediction Time Point Gate - to - Gate Rolling Airtime Rolling Departure Takeoff Landing Arrival Accuracy of existing prediction Avg. deviation estimated vs. actual OnBlock 17 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
How? Unleashing The Power of Big Data Historic Data Real-Time Data Historized Flight Data Stream Predicted Eurocontrol Arrival Time Data Warehouse Reporting Server Weather Data Stream 18 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
The Cycle 19 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
Parameters for the Model Delay @ Takeoff Season Lagged features Weight Weather Flight Duration (en route & @destination) Route, distance , … Traffic situation Runway(s) Passenger data #Passengers with connection % occupied seats 20 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
Parameters for the Model Delay @ Takeoff Season Lagged features Special disruptions Weight Weather Flight Duration (en route & @destination) Actual data needed Route, distance , … Traffic situation Capacity index Runway(s) Passenger data & airport information #Passengers with connection % occupied seats still missing 21 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
The Cycle 22 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
Model Building A few rolling window features are good! Prefer models that can be trained fast: Play around with features instead of models. 23 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
The Cycle 24 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
Visualization: Performance Monitor Data Made Up (No Actual Data)! Error @ flight done % Error @ minutes before arrival 100 60 90 50 80 70 40 60 Error Error 50 30 40 20 30 20 10 10 0 0 100 90 80 70 60 50 40 30 20 10 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% other prediction our prediction other prediction our prediction 25 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
How To Save Money? Gate Z50 Gate Z99 26 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
How To Save Money? Gate Z50 Gate A1 Gate Z99 27 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
How To Save Money? Gate Z50 Gate A1 Gate Z99 28 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
Visualization: Benefits Monitor Data Made Up (No Actual Data)! Potential benefit (k € ) 7 5 Potential Benefit 1000* € 6 4 #Saved Passengers 5 3 4 3 2 2 1 1 0 0 Saved Passengers Potential Benefit 29 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
The Cycle 30 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
Automation: Supervisor Supervision of Model Runs every hour Asserts that … enough data enough predictions predictions produced fast not too much missing data … 31 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
Lessons Learned Go To The Real World Fast – Make Everybody Understand The Benefits Supervisor & dashboard as early as possible Business needs to understand the real impact Translate Data Science to business language Need for standards Prototype needs to be translated to real application You do not want to rebuild everything 32 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
Partnership Lufthansa Industry Solutions | RapidMiner Partnership founded in 2015 RM widely used in Lufthansa Keeps up with market technologies For prototyping and productive applications Still lacks some software engineering capabilities … We pair it up with Hadoop, Oracle, Teradata , R, Python, … Lufthansa Industry Solutions: certified RapidMiner trainers 33 9/1/2017 From Prototype to Operative - Data Analytics @ Lufthansa
Many Thanks for Your Attention!
Recommend
More recommend
