African cow breeds with distinctive adaptation features and productivity potential include Kuri, N'Dama, and Bororo. These breeds represent unique genetic resources. Conservation and genetic advancement depend on an understanding of their genetic diversity and population structure. In this work, we examined genetic diversity and breed classification using high-density SNP genotyping data. We reported strong genetic differentiation due to extremely high allele frequencies (delta > 0.8) revealing highly informative loci for breed discrimination. Random forest classification identified twenty highly informative SNPs for breed differentiation and revealed traceability and assignment of the breeds. Confusion matrix revealed variation in breed assignment power. The cattle, Kuri, N’Dama and Bororo were perfectly classified indicating unique genetic markers peculiar to the populations. It is therefore concluded that, random forest, a machine learning approach and breed informative markers could be used for breed traceability, assignment and assessment of genetic diversity meant for cattle breed conservation and improvement.