In this study, we used a sample of 338 galaxies ---within the redshift range of $ 0.02<z<0.1$ drawn from the Sloan Digital Sky Survey (SDSS)--- for which there are available classifications based on their emission line ratios. We identified and selected Compton-thick (CT) AGN through the use of X-ray and infrared luminosities at $12\ m$. We constructed the spectral energy distributions (SEDs) for all sources and fit them using the CIGALE code to derive properties related to both the AGN and host galaxies. Employing stringent criteria to ensure the reliability of SED measurements, our final sample comprises 14 CT AGN, 118 Seyfert 2 (Sy2), 82 composite, and 124 low-ionization nuclear emission-line regions (LINER) galaxies. Our analysis reveals that, irrespective of their classification, the majority of the sources lie below the star-forming main sequence (MS). Additionally, a lower level of AGN activity is associated with a closer positioning to the MS. Using the D$_n$4000 spectral index as a proxy for the age of stellar populations, we observe that, compared to other AGN classes, LINERs exhibit the oldest stellar populations. Conversely, CT sources are situated in galaxies with the youngest stellar populations. Furthermore, LINER and composite galaxies tend to show the lowest accretion efficiency, while CT AGN, on average, display the most efficient accretion among the four AGN populations. Our findings are consistent with a scenario in which the different AGN populations might not originate from the same AGN activity burst. Early triggers in gas-rich environments can create high-accretion-rate supermassive black holes (SMBHs), leading to a progression from CT to Sy2, while later triggers in gas-poor stages result in low-accretion-rate SMBHs like those found in LINERs.