We design and develop AI solutions that align with your organization’s goals.
We use them to uncover complex patterns, predict developments, and reconstruct missing information. This helps you make better-informed decisions and improve performance.
In close collaboration, we develop a solution that fits your challenge, context, and way of working. We explore the available data, assess what AI can achieve, and identify which outputs will be useful in practice.
We combine your data with relevant data sources and develop models that integrate smoothly into your existing processes. For example, we connect the outputs to your BI solution and add the enriched information to your dashboards.
We use AI to solve specific challenges, including those in public transport. From estimating missing destinations and predicting running times to providing insight into expected crowding.
Enrich your check-in-only data with estimated check-out data
In public transport networks where passengers only check in, their boarding stop is recorded, but their alighting stop remains unknown. This makes it difficult to gain insights into passenger flows and vehicle occupancy.
Our OD Generator uses AI to estimate the alighting stop for each check-in. The solution combines boarding data with information about the transport network and the surrounding areas.
By enriching boarding data, our trained AI model provides insights into passenger movements, transfers, vehicle occupancy, and passenger kilometers. This helps operators better align their services with passenger demand.
How does it work? You send us your check-in-only transaction data, and within a few days we return the dataset enriched with estimated alighting stops. You can then use the data directly or explore the results in the interactive dashboard included.
Predicting trip times for new routes and diversions
New routes and diversions often lack historical running-time data. Yet planners need reliable estimates to create realistic timetables.
Our AI solution predicts running times between consecutive stops. The model combines historical trip data with route characteristics, weather conditions, and the timing of each trip. This enables it to provide reliable running-time estimates for route segments where no historical trip data is available.
This helps operators plan and compare alternative routes quickly and reliably. The AI model improves planning quality while reducing planning time and costs.
Predicting passenger crowding in public transport
How crowded will it be tomorrow? Our AI-powered crowding indicator predicts passenger occupancy and translates these predictions into clear crowding levels.
The solution makes occupancy predictions easy to integrate into passenger information systems. By providing insights into expected crowding, it helps passengers make informed decisions about which service to take.
Compared with measured occupancy, the predicted distribution across crowding categories matches the actual distribution by 93.8%, compared with 83.9% for the existing indicator.
We start by clarifying goals, questions, and context. Which decision or process needs support? What data is available, and what is missing? Together, we identify where AI can and should add value.
We design the data approach, the model, and how users will work with its outputs. We define the required accuracy and how to evaluate performance. Where appropriate, we start with a proof of concept to assess feasibility.
We bring together the relevant data and develop, train, and test the model. We compare its outputs with real-world observations and refine the model. Once the results meet the application’s requirements, we integrate it into the relevant systems and processes.
We support users in applying the outputs and monitor the model’s performance. As data or conditions change, we help adapt the solution. This keeps the application aligned with day-to-day operations.
An AI solution only has an impact when people understand its outputs and know how to use them. We guide and train users to interpret predictions, recognize uncertainty, and apply the insights.
This helps teams use AI responsibly and independently in planning, analysis, and decision-making.
Copyright © Zight. All rights reserved.