This research work proposes a low-cost and effective food spoilage detection by using multi-sensory setup by using a list of gas sensors . In recent years there has been a rapid development of this method, especially in the area of food control. The work has to its credit a mechanism that performs dimensionality reduction based on decision boundary and logistic regression. The system trains different models using high quality training sets, test sets, and feature combinations for classification which serve as the basis for obtaining more accurate results. The classifier engine is responsible for classifying food items into one of the two classes: “spoiled” or “fresh”. The experimentation results clearly reveal the capability of the system to produce accurate results still maintaining low cost and versatility of usage.