Predict a number of your passengers and bags for each flight in 3 simple steps
Why do I need pax flow forecasting?
Airport operating costs depend largely on the flow of passengers and baggage. At the same time, the flow is extremely uneven and depends on the season, day of the week, time and airlines. An accurate forecast will allow you to optimize your operating costs.
Optimize Flight Schedule
Optimize your Flight Schedule to avoid unnecessary incidents
Plan & Optimize Airport Resources
Counters, Staff, Vehicles, Infrastructure, Fuel, etc.
Estimate your spending
Estimating expenses allows you to avoid unplanned expenses and reduce costs
Make "What If" analyze
Close/open Counters, Attract new airlines, Make changes to Schedule,Open new routes, Change Gateways, and Change procedures.
How does it work?
4 simple steps to run ML for passenger prediction
1. Load your data
Export XLS file with flight history from your AODB. It is necessary to export number of passengers and bags. Than load this file to the system.
2. Automatically train your model
The system will analyze a history of your flights and automatically find some patterns. It is called AutoML because this action doesn't require any actions from the user. It is fully automatically.
3. Load the future schedule to predict
You just load XLS with future flight schedule to the system and get result with predicted number of passengers and bags.
4. Compare results with actual data
You can compare results of prediction with your actual data to ensure an accuracy of the system.
4 Facts about our AutoML service
3 Features
Passengers
Bags
Bags weight
High accuracy
Accuracy of prediction is very high. Mean error is less than 5%.
Very Simple
User just works with MS Excel XLS file. No knowledge of ML is needed.
15 minutes
is needed to get your 1st prediction
Features of prediction
What you can predict with this service
Passengers (all tiers)
Transfer passengers (coming soon)
Bags (Pro+ tier)
Fuel (coming soon)
Bags weight (Pro+ tier)
Flight delay (coming soon)
Accuracy of prediction compared to actual data
A couple words about accuracy
It totally depends on your history data. More data you load - more accurate it becames.
Short term (1-2 months) accuracy typically is under 5%. Long term (about 1 year) accuracy is about 5-10%, because it depends on macro economic factors.
Why Choose our Software services?
It's SIMPLE!
You don't need to keep high qualified engineers to build high quality model. It can be done by current specialists.
It's FAST!
It works at the speed of an plane.
Get the 1st results in 1 hour!
A look into the future
You see a loading of your terminal facilities for days and weeks in future.
Happy passengers and airlines
You manage your airport proactively and provide the highest level of Quality
Who needs our services at the airport?
Operational team
Streamline operations and enhance efficiency. Optimize the flight schedule.
Financial team
Make informed financial decisions based on predictive analytics
Ground Handling team
Optimize staff count for serving passengers and baggage.
Quality control team
Improve service quality and customer satisfaction.
Commercial team
Drive revenue growth through better resource allocation and passengers satisfaction.
IT team
Enhance existing AODB & RMS systems through integration with our cloud AI&ML services.
Frequently Asked Questions
Question:
What kind of specialist do we need to run it?
Answer:
We have made this AutoML service as simple as possible. Basic knowledge of Excel is enough to use it. Thus, any airport employee can start using this machine learning service.
Question:
I have tried it, but accuracy is very low. What can I do?
Answer:
Please contact our technical support team. We will review your data files and will get some recommendations to improve prediction.
Question:
What is about privacy of my data files?
Answer:
Your data will be stored in private cloud storage and will be available only for you. We don't use your data for anybody else. Model trained on your data will works with your data only.
Question:
What technologies are under the hood?
Answer:
We use the CatBoost library for gradient boosting on decision trees. Before training the model, more than 30 features are added to the data tables.