Assuming you have proper data (no charging effects etc.), that's pretty straightforward: you check out a database, e.g. NIST XPS: https://srdata.nist.gov/xps/EngElmSrchQuery.aspx?EType=PE&CSOpt=Retri_ex_dat&Elm=Fe
As you can see, e.g. the 2p3/2 peak would be at 711eV for Fe2O3 or 709-710eV for FeO. If you have a mixture, you may require to do a deconvolution which is more tricky, you can read about it in the CasaXPS manual which is helpful even when you don't have CasaXPS: http://www.casaxps.com/help_manual/manual_updates/peak_fitting_in_xps.pdf
If you're are expecting a single oxide type, then you can tell this form the shape, binding energy and satellite positions.
If you have mixed valence states, then fitting would be required for actual percentages, which for iron is rather difficult due to the multiplet structure. That said, you can take a more generalized approach in those cases as say "the sample contains Fe(II) and Fe(III) species"
A good reference is: https://www.thermofisher.com/uk/en/home/materials-science/learning-center/periodic-table/transition-metal/iron.html
Ashwani Ashwani My experience of such systems is that the surface (top 10 - 15 atomic layers maximum) is always fully oxidized (i.e. all would be Fe3+ in your system from peak shift). The bulk Fe2+ can only be shown by (say) Ar+ etching of the surface to remove the top layers (including C) and get into the bulk composition.