Plastic is a ubiquitous material used in a wide range of applications, but its disposal has become a significant environmental concern. Bioplastics, which offer several advantages over conventional plastics, including a smaller carbon footprint, greater biodegradability, and adaptability, are being explored as a potential alternative. Among the bioplastics, starch-based polymers have gained significant attention due to their natural abundance, biocompatibility, low cost, and renewability. In this study, a starch-based bioplastic was developed using glycerol and sorbitol as plasticizers and wood dust as a filler through the casting method. Starch was extracted from sweet potato (Ipomoea batatas). The optimal conditions for preparing the bioplastic with minimum water adsorption were determined using a three-level Box-Behnken design. The amount of starch, plasticizer, and filler were the key parameters for this study. The percentage of water absorption was determined by immersing the prepared bioplastic in distilled water for 24 hours and weighing it. The result of the Fourier Transform Infrared (FTIR) spectroscopy of extracted starch showed a similar observation peak pattern when compared to the FTIR spectrum of commercial starch. The linear response model produced a good coefficient of determination (R2 = 0.8237), indicating that the chosen parameters implicitly affect the percentage of water absorption. The prediction percentage of water absorption after optimization is 20.069 with a desirability value of 0.806. This study shows that the response surface approach helps in determining the optimum conditions for the minimum percentage of water adsorption from starch-based bioplastic.