Case Study

University of Houston: MatLearn

In a joint collaborative effort, LaPraim and the University of Houston worked together to build a robust public site. The focus of attention was to create a website that combines machine learning capabilities with an easy-to-use web application.

Explore the site
University of Houston chemistry student working in the laboratory.

Problem Statement

Integration of Data-driven Results and Non-organic Materials

The challenge was to integrate data-based and experimental results to create functional non-organic materials. The objective of LaPraim was to develop a dedicated UI/UX in order to integrate the newly developed machine learning code into a web application.

The Solution

Machine Learning and Clear Functionality

Throughout the development of the project, the project heads utilized Flask, Python, scikit-learn, and Matplotlib. The use of Matplotlib, however, was predominant to opt for smooth charting functionality. Conversely, scikit-learn became crucial for the project to input machine learning data into the charts.

Main Framework

Moreover, Flask, the web framework, served as a framework for Python. Since it is a micro-framework, there was no need to use additional tools. Of course, the project developers used CSS, HTML, and JavaScript to ensure the developed web pages are interactive and presentable. The project heads also used a modern hosting solution like AWS rather than a traditional web hosting service. It made user understandability and maintenance easier.

Account manager Mali Gorovoy (Left) and full stack developer Keith Pittman (Right) holding element models at the Houston University campus

The Results

MatLearn: Predict Accurate Energy Formations

MatLearn: Predict Accurate Energy Formations

MatLearn essentially helps you predict the formation of energy. It’s the integrated machine learning feature that makes it possible to make compositions of diagrams and take play around with synthetic chemistry.

When it comes to compositional space, MatLearn is an ideal machine learning and visualization web-based model to explore a wide range of material properties. Ultimately, MatLearn guides SSS (solid-state-synthetic) to certain regions of a predicted diagram. It allows you to focus on stable compounds with ideal properties.

MatLearn Uses

The mechanics of MatLearn revolve around a combination of ternary and binary systems on the home page. It allows you to establish the right composition range and then create predictions. One of the wonders of MatLearn is that it immediately creates a predicted ternary or binary diagram

The average MatLearn estimation of binary systems is represented through the plotted blue dot. The shaded gray indicates the accuracy of predictions. Similarly, the compounds of training data are represented as vertical lines. In retrospect, it is a straightforward process to create accurate predictions and analyze data effortlessly.

Connect in 3 simple steps

Let's connect.

We want to hear from you and learn how we can make technology work for your organization.