Multi-component polymer systems are of interest in organic photovoltaic and\ndrug delivery applications, among others where diverse morphologies influence\nperformance. An improved understanding of morphology classification, driven by\ncomposition-informed prediction tools, will aid polymer engineering practice.\nWe use a modified Cahn-Hilliard model to simulate polymer precipitation. Such\nphysics-based models require high-performance computations that prevent rapid\nprototyping and iteration in engineering settings. To reduce the required\ncomputational costs, we apply machine learning techniques for clustering and\nconsequent prediction of the simulated polymer blend images in conjunction with\nsimulations. Integrating ML and simulations in such a manner reduces the number\nof simulations needed to map out the morphology of polymer blends as a function\nof input parameters and also generates a data set which can be used by others\nto this end. We explore dimensionality reduction, via principal component\nanalysis and autoencoder techniques, and analyse the resulting morphology\nclusters. Supervised machine learning using Gaussian process classification was\nsubsequently used to predict morphology clusters according to species molar\nfraction and interaction parameter inputs. Manual pattern clustering yielded\nthe best results, but machine learning techniques were able to predict the\nmorphology of polymer blends with $\\geq$ 90 $\\%$ accuracy.\n