Case Study: Matmatch improves materials data quality with Intellegens Alchemite™

Improving materials data quality

Matmatch, the materials search platform, needed a better way to work with its sparse materials databases, where many material properties were missing for ceramics, polymers, aluminium, and metals. The challenge was to improve data quality for material selection, help identify outliers, and make it easier for users to find suitable materials even when key properties had not been measured. Intellegens’ Alchemite™ machine learning toolkit was used for this purpose.

Intellegens implemented Alchemite™ to impute missing values and verify existing data across highly sparse datasets, integrating via API with Matmatch’s systems. The results were strong: on a metals dataset with 55,000 rows, 254 inputs, 162 outputs, and 96% missing target data, the model achieved an R² of 0.82 against key targets. Matmatch also reported that the tool was easy to use and outperformed expectations, enabling accurate prediction of twice as many material properties as originally planned.

Matmatch

Melissa Albeck

Chief Executive Officer