Decomposing chemical interactions into bonds and other higher order interactions is a common practice to analyse chemical structures, and gave birth to many chemical concepts, despite the fact that the decomposition itself might be subjective in nature. Fragment molecular orbitals (FMOs) offer a more rigorous alternative to such intuition, but might be less interpretable due to extensive delocalisation in FMOs. Inspired by the Principal Component Analysis in statistics, we hereby present a novel framework, Principal Interacting Orbital (PIO) analysis, that can very quickly identify the "dominant interacting orbitals" that are semi-localised and easily interpretable, while still maintaining mathematical rigor. Many chemical concepts that are often taken for granted, but could not be easily inferred from other computational techniques like FMO analysis, can now be visualised as PIOs. We have also illustrated, through various examples covering both organic and inorganic chemistry, how PIO analysis could help us pinpoint subtle features that might play determining roles in bonding and reactions.