The research is applied to 20 compounds contained into <em>Avicennia marina</em> which is : <em>4’,5-dihydroxy-3’,5’,7-trimethoxyflavone, isorhamnetin 3-O-rutinoside, quercetin, stenocarproquinone B, avicennone C, avicennone E, avicennone F, avicennone D, 4’,5-dihydroxy-3’,7-dimethoxyflavone, chrysoeriol 7-O-glucoside, naphta(1,2-b)furan-4,5-dione, 3-hydroxy-naphta(1,2-b)furan-4,5-dione, kaempferol, 5-hydroxy-4’,7-dimethoxyflavone, avicequinone C, 2-[2’-(2’-hydroxy)propyl]-naphta[1,2-b]furan-4,5-dione, 4’,5,7-trihydroxyflavone, luteolin 7-O-methylether, luteolin 7-O-methylether 3’-O-beta-D-glucoside, 5,7-dihydroxy-3’,4’,5’-trimethoxyflavone</em> which are expected to have an effect as antidiabetics agent. It is undertaken to find out the linier or nonlinier relationship of the structure activty (QSAR), the physicochemical characters (lipophilic, electronic, dan steric) to the activity changing (<em>rerank score</em>). Moreover, it is also undertaken to find out which one of those 20 <em>Avicennia marina </em>compounds are predicted to possess the activity as the best antidiabetics agent by the comparison of pioglitazon. It is arranged through computerized methods or <em>in silico</em> not only for having the structure of two and three dimensions but also knowing the physicochemical characters too; through some programs such as <em>ChemBioDraw Ultra </em>12.0 and <em>ChemBio3D Ultra </em>12.0 from <em>CambridgeSoft. The docking </em>program is used to make prediction of the activity by using <em>Molegro Virtual Docker </em>2011.5.0.0 from CLCBio<em>.</em> by using the <em>peroxisome proliferator-activated receptor-gamma </em>(PPAR𝛾) in code of PDB: 2XKW. The result of the regression analysis through IBM SPSS program shows nonlinear relationship among structures, <em>Avicennia marina</em> physicochemical characters of twenty compound contents dealing with the activity changing. The compound of <em>isorhamnetin 3-O-rutinoside</em><em>, chrysoeriol 7-O-glucoside, </em>and <em>luteolin 7-O-methylether 3’-O-beta-D-glucoside </em>are predicted to have the best antidiabetics agent if it is compared with the comparing drug pioglitazon.