Since you say "admet for a specific compound", I assume you are looking for experimentally determined values for just one compound. That should be fairly easy. The only way to get accurate experimental ADMET data is to refer to the actual published study. You need to look up the paper or patent document where the ADMET studies for this compound have been reported, by performing a pubmed and patent database search using appropriate keywords. IF you are unable to find it in literature then it is quite possible that it doesn't exist. IT is not necessary that ADMET properties may have been determined for a particular compound experimentally. Usually if such a study is done then the values are populated in popular databases like Pubchem, DrugBank etc. You can also check ZINC database. IF this compound is part of a well known class of drugs/inhibitors, you can check in related papers with studies on QSAR or pharmacophore modelling, where this molecule may have been used as part of a control or training dataset. In that case, the ADMET properties or potential references for it would be listed there. IF you need such values for a large number of compounds, manual literature search will become tricky and laborious so you may have to rely on the information present in databases. Also, predicted/in silico values can never be used as a substitute for experimental data for control dataset as is required in your planned project. IF you can't find the ADMET properties for one particular molecule, you may exclude it from your dataset and use others. There are 1000s of molecules with published SAR, QSAR, pharmacodynamics and pharmcokinetics data published in literature and patents.

As Aditya states, you must scour the literature to find these data, if it exists. While I have no experience with Article admetSAR: A Comprehensive Source and Free Tool for Assessmen...