Principal Component Analysis
Principal Component Analysis (PCA) is a useful statistical technique for identifying patterns in large datasets, and expressing the data in such a way as to highlight their similarities and differences.
It can take large, multivariate datasets and convert these into more manageable tables of data, identifying new and potentially meaningful underlying variables. At PMCC we have used PCA to generate proprietary solvent maps which we use for simple yet effective solvent screening but also in combination with DoE to assess the affect of solvent on a reaction. This can bring order to investigating the way in which different chemical components affect a given reaction and enables us to investigate discrete factors (i.e. solvent) by their chemical properties (or descriptors), allowing the relation of solvents, which is not normally possible. We can provide our clients with solvent selections from our PCA maps, assist with screening a wide range of solvents to efficiently understand the chemical space, and combine solvent screens with the screening of physical parameters e.g. temperature, stoichiometry, concentration etc.
The combination of DoE and PCA facilitates the development of a reaction by informed decision making, where all critical factors may be considering alongside one another in an efficient and effective manner. PCA has also been applied to ligand databases with our collaborators, enabling rational catalyst and ligand screening.
Contact us to discuss how we can work together get in touch
For examples of combining Design of Experiments and Principal Component Analysis in catalysis see the following case studies:
- Case Study 1: Development of a challenging Suzuki reaction.
- Case Study 2: Development of a Buchwald Hartwig amination reaction.
- Case Study 4: Optimisation of a Suzuki–Miyaura reaction through solvent selection.