I have performed the optimization of hexapyrrole for determination of raman spectra using B3LYP geometry with 631G(d) as basis set in gaussian. However I am unable to extract the raman spectral data in order to plot it in origin software.
You can use Multiwfn(http://sobereva.com/multiwfn/) software by Prof. Tian Lu (http://www.keinsci.com/members/Tian_Lu_EN.shtml) for easy plotting of Raman Spectra. Please check the recent Manual (Section 3.13 for instructions and 4.11.5 for worked out example)
You can extract the data If you have access to Gaussview, after opening the output file(.log or .out) by Gview, Go to Results>Vibrations. A new window will open. Click the "spectra" button at the bottom of this page and you will see the RAMAN spectra along with other spectra (Please ignore this part if you've already done). now right click on the Raman spectrum and "save data" to a .txt file . You can use this file to plot with any external software (Tested with Gaussview 6).
It is worth to note that Gaussian itself only outputs Raman activities, however, in order to plot the spectrum that can be comparable with experimental Raman spectrum, the Raman activities must be first converted to Raman intensities, otherwise the height of the peaks will be unreasonable.
The spectrum plotting interface of Multiwfn, which has been mentioned by Turbasu Sengupta, could convert the Raman activities loaded from Gaussian to intensities and then plot corresponding Raman spectrum. GaussView is also able to plot Raman spectrum, however, the spectrum is directly plotted based on Raman activities, and thus not as strict as the one simulated by Multiwfn.
In the spectrum plotting module of Multiwfn, if you want to export raw data of Gaussian, you can select option "-1 Show transition data"; while if you want to export curve points of the simulated spectrum, you can select "2 Export X-Y data set of lines and curves to plain text file", then you can directly import the .txt file to Origin and redraw the spectrum.
Can I also use gauss sum for data extraction as it has an option to scale up the values by 0.97. I have read in some papers that raman values has to be scanned by 0.97.
The spectrum plotting module of Multiwfn is very flexible, of course it provided an option used to set scale factor (i.e. "14 Multiply the vibrational frequencies by a factor"). Furthermore, in Multiwfn, the scale factor can not only be applied to all frequencies, but can also be applied to specific frequencies (e.g. those higher than 500 cm^-1).