The Parking Spot: AI-Powered Demand Forecasting Across 70+ Locations

Client Overview

The client is a U.S.-based provider of near-airport parking, operating more than 70 locations across multiple states. Customers book services through web and mobile applications, and the business continues to scale rapidly across new locations and demand profiles.

The Challenge: Manual Forecasting Can’t Keep Up with 70+ Locations

Analysts previously relied on manual, spreadsheet-based estimates — reviewing booking histories, holidays, and travel trends — to set prices and plan shuttle schedules. As the business expanded beyond 100 product offerings, this manual approach could no longer keep pace with demand variability, creating the need for an AI-powered forecasting solution capable of forecasting demand up to 30 days ahead across every site.

The Solution: A Multi-Model Forecasting Service Built on Azure

DataArt led an eight-month demand forecasting consulting engagement from data discovery and proof of concept to an MVP forecasting service on Azure. The solution forecasts daily demand up to 30 days ahead and feeds the client's revenue management system to support pricing and staffing decisions.

Multiple models were evaluated during the proof-of-concept phase. The team selected and productionized a multi-model approach using Chronos, with continuous training and monitoring. The forecasting service integrates seamlessly into the client's revenue management system, allowing analysts to use forecasts directly in pricing and planning workflows.

The Impact: 30-Day Visibility Embedded in Revenue Decisions

Key Advantages of Machine Learning Demand Forecasting Consulting

• Replaces Prophet forecasts with a repeatable, governed process that scales with new sites and market changes.
• Combines internal bookings and stays with a variety of additional data sources for more reliable demand signals, forming the basis of airport car parking analytics software.
• Integrates with existing revenue management tooling so analysts can adjust prices and staffing based on forward-looking demand.
• Creates a data foundation for new use cases such as labor planning and shuttle routing driven by the same forecasts.

Technology Snapshot

Cloud

Microsoft Azure

Data and Modeling

Python-based model, Chronos for continuous training and monitoring.

Data Sources

Internal bookings and stays paired with a variety of additional data sources.

Integration

Backend of the client's revenue management system, supporting pricing and operational planning.

Why DataArt?

Long-term technology partner for the client's core IT landscape, ensuring faster alignment and smoother communication across teams.

DataArt’s Azure machine learning consulting team brings direct experience building production-grade forecasting and revenue management solutions in Azure with Python-based models.

Ability to adapt the team composition (data engineering, MLOps, DevOps) as the project moved from PoC to MVP and support.