In recent years, the field of electrochemistry has seen remarkable progress, drawing a considerable attention to the potential of electrochemical impedance spectroscopy (EIS) in various domains, particularly in agriculture where ensuring the quality of fruits and vegetables is essential to the production process. EIS stands out as an accessible, and reliable technique, made feasible with compact and cost-effective hardware for measuring spectral impedance. Nonetheless, the existing impedance measurement systems primarily rely on the conventional frequency sweep method, which can be time-consuming. This has prompted the demand for faster systems better suited for real-time measurement and rapid detection. This research adapts a novel set of wide-band signals, previously unexplored in impedance measurement, coupled with a suitable machine learning data processing tool. This innovative approach facilitates the accurate classification of impedance spectra, encompassing information tailored to the specific application at hand, thereby enhancing the versatility of our methodology. The wide-band signals demonstrate not only a relatively flat power spectral magnitude but also a controllable time domain maximum amplitude, enabling the effective measurement of various types of bio-samples. Experimental results are presented to verify their accuracy, comparing them to a standard precision research-grade impedance analyzer. Notably, the classification process encompasses the frequency range of sub-Hz, below 1 Hz, a distinctive aspect that distinguishes our research from prior studies. This inclusion broadens the scope of our investigation and contributes to a more comprehensive understanding of impedance characteristics. Subsequently, machine learning-based prediction models are developed, trained, and tested using information extracted from EIS spectra. These models showcase an exceptional ability to precisely predict the properties of new samples, providing an alternative, real-time, on-field estimation of the aging condi