ENEOS Materials: Driving Material Innovation With Predictive Modeling

ENEOS Materials is a global leader in the research, development, and manufacturing of rubber and elastomers. In their tire materials development department, they not only develop high-performance rubber materials used for tires, but also provide technical support by proposing optimal material formulations that meet unique performance specifications that customers need.

Because tire rubber compounds consist of a wide variety of materials, the formulation process is highly complex. This means the knowledge accumulated by individuals through years of experience becomes extremely important.

The researchers at ENEOS Materials faced delays in responding to customer service (and missed opportunities for product sales and adoption) because finding the best solutions can take a significant amount of time.

They needed a solution that would allow them to quickly propose formulations that meet customer requirements.

Meeting Customer Needs With Accurate Formulation Plans

ENEOS Materials researchers had a simple goal in mind: Design a system that could propose highly accurate formulation plans that meet customer requirements, regardless of the researchers’ experience. It needed to be user-friendly and accessible to a range of teams in the organization. The main issue was that, because of a shortage of specialized data-focused team members, the R&D team had to develop the system themselves with significantly less data expertise.

In partnering with Dataiku, ENEOS Materials was able to solve many complex foundational challenges:

• The original datasets the team used contained missing values and schema inconsistencies, which made them unusable for building machine learning (ML) models. They took advantage of Dataiku’s tools for visualizing dependencies and its flexible visual recipes and, through this data preprocessing, were able to easily construct datasets that were well-suited for ML.
• ENEOS Materials could easily interpret model performance from multiple perspectives with ML algorithms and explainable AI features available on Dataiku. This enabled ENEOS Materials to obtain the best models for predicting compound properties from formulation information without writing any code, deepening their understanding of complex formulations.
• Through collaboration with Dataiku and use of Python recipes, ENEOS Materials was able to validate different algorithms and constraints. From here, they developed an algorithm that was able to propose diverse and feasible formulation conditions without relying on business experience. In addition, they also generated innovative formulation proposals that were difficult to conceive with traditional methods.
• To build the UI and ensure a user-friendly application, they used Dataiku scenarios and Dataiku applications. Users were able to provide input in its design, contributing familiar terms and optimal formulation plans tailored to their uses.

Now that ENEOS Materials has a strong data science foundation, they’ve also started to explore different use cases and ways to expand their ML and AI practice.

Increased Efficiency and Production Opportunities

Since implementing their unique ML solutions, ENEOS Materials has taken advantage of improvements in their operational efficiency, as tasks that were previously handled manually — data preprocessing, analysis processes, and planning — are now automated. This has led to a significant reduction in the time needed for data preparation and analysis, which means researchers can allocate time to more value-add tasks. ENEOS Materials is also utilizing DevOps to update the models while also advancing operations in the field.

In the future, we expect the speed of customer proposals to improve significantly, reducing the timeline from the previous span of several months to just a few days. Takumi Adachi Manager at ENEOS Materials

Specifically, ENEOS Materials anticipates that this use case will generate value in two specific areas.

1. Reduction in evaluation effort: Researchers can obtain results in much less time, potentially improving the efficiency of the entire development process. The evaluation process previously took several weeks, which they now expect to be completed in a few days. Consequently, this will lead to a more effective use of resources, allowing their team to focus on other, more critical business initiatives.
2. Improved customer response speed and increased product adoption opportunities: In situations where rapid response is important, the speed the team can leverage from proposal to adoption can potentially contribute to increased sales.

Dataiku has provided immense value through its rich features and user-friendly interface. Taku Ichibayashi Group Manager at ENEOS Materials

Foundationally, the team at ENEOS Materials has not only been able to solve for their use case, but to build a growing team of citizen data scientists in their organization. Beginners were easily able to handle data analysis and predictive modeling, and they were also able to hone their skills in the Dataiku Academy. If they found themselves in need of help, Dataiku’s technical support helped resolve issues quickly, preventing delays.

ENEOS Materials’ journey in ML and AI showcases how even simple implementation can help solve complex industry challenges. Partnering with Dataiku streamlined their formulation process, boosted efficiency, and empowered teams to embrace data science, all while fostering ongoing innovation.