Pengujian array sensor berbasis membran lipid untuk klasifikasi pola rasa kebasian susu menggunakan metode Linear Discriminant Analysis (LDA): Studi kasus pada susu sapi dan susu kambing

INDONESIA:

Telah dilakukan pengujian sistem array sensor berbasis membran lipid yang digunakan untuk klasifikasi pola rasa perkembangan kebasian susu sapi dan susu kambing. Array sensor dibentuk dari 16 membran polimer sebagai elektroda selektif ion (ISE). Pengujian dilakukan dari kondisi segar sampai kondisi basi selama 12 jam. Respon sensor dievaluasi menggunakan metode pengenalan pola Linear Discriminant Analysis (LDA).

Pengolahan data LDA mengguakan metode stepwise untuk menyeleksi variabel diskriminator yang memiliki kontribusi dalam prediksi pada model fungsi diskriminan. Berdasarkan fungsi diskriminan ke-1 dengan nilai varian 99.8 % pada susu sapi dan nilai varian 100% pada susu kambing dapat membedakan dan mengklasifikasi perkembangan kondisi rasa basi pada susu sapi maupun susu kambing.Hasil pengolahan LDA diperoleh ketepatan klasifikasi pada susu sapi 100% dari kondisi data awal dan 100% dari metode “leave one-out” cross validation. Sedangkan pada susu kambing diperoleh 96.5% ketepatan klasifikasi pada kondisi data awal dan 94.5% dari “leave one-out” cross validation.

ENGLISH:

Testing array system sensor based on lipid membrane had been conducted and it was used to classify taste pattern of cow and goat milk staleness. Array sensor is constructed from 16 polymer membranes as ion selective electrode (ISE). The tasting was done from the fresh condition up to stale condition for 12 hours. The sensor response was evaluated by using pattern recognition method Linear Discriminant Analysis (LDA)

The data analysis of LDA used stepwise method to select discriminator variable that has a contribution in predicting discriminant function model. Based on the first discriminant function with 99.8% variant value in cow milk and 100% variant value in goat milk, it can differentiate and classify the development of the stale condition either in cow milk and goat milk. From the result of the LDA analysis, it was found that the constancy classification for cow milk is 100% from the original data condition, and 100% from “leave one-out”cross validation. On the other hand, 96.5% was found in goat milk from the original data condition and 94.5% from “leave one-out” cross validation