Coyote: From Churn Analysis to Predictive Safety

Coyote uses IoT-based devices and mobile applications that enable their users to warn other drivers of traffic hazards. Having started out with predictive analytics for churn prevention, today Dataiku has become an integral part of Coyote’s predictive safety operations.

In order to successfully implement a data-driven strategy and embark on the journey to Enterprise AI, companies often need to start small with simpler, concrete impact projects, in order to convince stakeholders in the value of their data initiatives before moving on to more advanced applications. 

This is precisely what Coyote, the European leader in real-time road information and one of Dataiku’s oldest customers, was able to do. 

Coyote uses IoT-based devices and mobile applications that enable their users to warn other drivers of traffic hazards and conditions (e.g., traffic obstruction, accidents, speed cameras, etc.) that are detected while driving. Initially having started out with predictive analytics for improving their customer retention, today Dataiku has become an integral part of Coyote’s predictive safety operations.

See How IoT is Shifting Enterprise AI Strategy

Get the White Paper

Ensuring Subscriber Retention and Loyalty With Predictive Behavioral Analytics

Through its connected devices, Coyote collects extensive data on the different uses of its community, such as mileage, time spent on the road, or the number of alerts issued by the community members.

Before venturing into predictive safety for their core services, they built and implemented a predictive behavioral analysis application for customer knowledge and service improvement. The application automatically compiles and processes heterogeneous and completely anonymized data (contractual data, customer declared data, real-time device data…).

This data is then processed by a machine learning algorithm to model user behavior. This model and its results were subsequently adjusted in order to optimize marketing campaigns, and resulted in an 11% increase in efficiency of their outbound campaigns thanks to increased knowledge of the actual uses of the service.

By improving retention rates, Coyote wished to enhance the following virtuous circle: the more users they acquire and retain, the better the service quality of their applications, and vice versa. What’s more, having proven the value of leveraging predictive analytical capabilities through a centralized platform for their marketing operations, Coyote’s data team now had the opportunity to evangelize the use of data science and predictive machine learning for the company’s core services – using IoT devices to improve road safety.

In order to prove the DataLab’s value and evangelize Dataiku across the organization, we had to start off with a project that demonstrates concrete business impact and drives ROI, which is why we initially focused on customer retention. However, the idea from the very start was to leverage the value of our vast data sources and implement a data-driven strategy at the heart of our core products and services, which Dataiku allowed us to do rapidly and effectively.”

Florian Servaux, Head of DataLab, Coyote

Watch Video

Predictive Safety to Improve Roads

More recently, Coyote has developed a project using Dataiku to leverage vast amounts of IoT-derived data in order to identify steep, potentially dangerous turns on car roads and, based on that, to develop a dynamic recommended speed limit model to prevent road accidents. The machine learning model facilitated the detection of speed limit anomalies and, consequently, enabled Coyote to estimate the global quality and reliability of the displayed speed limit.

The project is comprised of four major elements:

  1. Identifying all S-curves in France and calculating their angle, as well as the distance between them;
  2. Developing a dynamic recommended speed limit model based on this data;
  3. Building and releasing a database of dangerous road curves;
  4. Monitoring and service optimization.

In order to achieve that, Coyote uses a multitude of diverse and complex data sources, including CRM data from their users’ activity, alerts about road accidents and other events, maps and GPS data, and data from other IoT devices and applications.

Watch Video

To ensure accuracy and efficiency, Coyote needed a centralized platform to connect to, clean, prep, and integrate all of these data sources. Dataiku provided the end-to-end solution for:

  • Connecting to data
  • Building the dynamic speed limit model
  • Feature engineering
  • Deployment of database
  • Data visualization and visual AutoML
  • Model monitoring and optimization

Dataiku’s centralized and elastic environment, collaboration features and focus on operationalization have allowed Coyote to significantly reduce the time needed for feature engineering, successfully deploy and continuously optimize their predictive maintenance products and services for delivering quality driver assistance and improving road safety for everyone.

Future-Proof Your Operations With Predictive Maintenance

Get the White Paper

Dataiku for Predictive Maintenance

Customers building predictive maintenance solutions with Dataiku benefit from:

  • The ability to centrally and seamlessly connect to data, wherever it’s stored.
  • A simple and fast interface for ETL, including interactive data cleaning and integrated advanced processors.
  • AutoML features, including the ability to compare dozens of algorithms directly from the  Dataiku interface ( both for supervised and unsupervised tasks).
  • One-click model deployment on the cloud with Kubernetes.
  • Robust model monitoring features to prevent model drift.

Predictive Maintenance in Transportation and Manufacturing

Predictive maintenance models can empower organizations to get ahead of system failures to tackle infrastructure repairs on their own terms. With Dataiku, predictive maintenance is even easier.

Read more

Transforming Predictive Maintenance with AI

When it comes to AI + predictive maintenance, manufacturing and many other sectors have a lot to gain. Rather than correcting problems once they occur, predictive maintenance prevents problems from ever happening in the first place.

Learn More

Anomaly Detection for the AI-Driven Business

Across industries and use cases, there is perhaps no other data science strategy more important to leverage than anomaly detection.

Learn More

The Evolving Role of IoT in the AI Revolution

Industrial use of internet of things (IoT) connected devices has been around for years. So what's the next big thing in the IoT space?

Learn More

The Automotive Industry: Driving the Future of AI

Automobile data analytics isn’t just about self-driving cars; data science and machine learning technologies can help keep auto organizations competitive by improving everything from research to design manufacturing to marketing processes.

Learn More