Résumé : [Abridged] JWST observations of the Orion Bar have shown the incredible richness of PAH bands and their variation on small scales. We aim to probe the photochemical evolution of PAHs across the key zones of the photodissociation region (PDR) that is the Orion Bar using unsupervised machine learning. We use NIRSpec and MIRI IFU data from the JWST ERS Program PDRs4All. We lever bisecting k-means clustering to generate detailed spatial maps of the spectral variability in several wavelength regions. We discuss the variations in the cluster profiles and connect them to the local physical conditions. We interpret these variations with respect to the key zones: the HII region, the atomic PDR zone, and the three dissociation fronts. The PAH emission exhibits spectral variation that depends strongly on spatial position in the PDR. We find the 8.6um band to behave differently than all other bands which vary systematically with one another. We find uniform variation in the 3.4-3.6um bands and 3.4/3.3 intensity ratio. We attribute the carrier of the 3.4-3.6um bands to a single side group attached to very similarly sized PAHs. Cluster profiles reveal a transition between characteristic profiles classes of the 11.2um feature from the atomic to the molecular PDR zone. We find the carriers of each of the profile classes to be independent, and reason the latter to be PAH clusters existing solely deep in the molecular PDR. Clustering also reveals a connection between the 11.2 and 6.2um bands; and that clusters generated from variation in the 10.9-11.63um region can be used to recover those in the 5.95-6.6um region. Clustering is a powerful tool for characterizing PAH variability on both spatial and spectral scales. For individual bands as well as global spectral behaviours, we find UV-processing to be the most important driver of the evolution of PAHs and their spectral signatures in the Orion Bar.