Crystal structure classification of binary intermetallic structures with 1:3 stoichiometry was done with simple machine learning algorithms. The successful crystal structure segregation is attributed to the novel set of descriptors comprising both compositional and structural features. The dataset includes 97 features, and a total of 2366 reported compounds adopting six different structure types. The unsupervised learning method based on principal component analysis (PCA) followed by clustering using the K-means method was applied to cluster compounds belonging to different structure types. Using the recommendation engine, we predicted the expansion of the clusters and then identified cluster/structure-type overlap. PuNi3-type was among the clearly segregated structure types according to the unsupervised model, and a novel representative, TbIr3, was selected for experimental validation, adopting this structure. The final supervised machine learning predictions were done with PLS-DA, SVM, and XGBoost confidently predicting the novel TbIr3 to belong to the PuNi3-type with an accuracy of 96.6%, 99.8%, and 99.9% respectively. Analysis of the features reveals that the main contributing features to the AB3 crystal structure segregation are the average shortest distance count of A element, the total number of sites of B atom, and the total number of second shortest distance count of B atom. Given that the predicted PuNi3-type of the TbIr3 phase could be controversial due to the extensive study of the Tb–Ir phase diagram and the reports of the TbIr3 in two different structure types, we conducted two independent experimental structural validations to confirm the existence of the TbIr3 in PuNi3-type structure. Subsequent theoretical validation explains that Ir-Ir contacts are the primary stability factor of TbIr3 in PuNi3-type structure over other structure types.