Lafnidita Farosanti

Analisis Data CATA Hasil Uji Sensori Produk Coklat Menggunakan Algoritma Naïve Bayes dan Tools XLSTAT

Introduction

Analisis data cata hasil uji sensori produk coklat menggunakan algoritma naïve bayes dan tools xlstat. Analisis data CATA uji sensori coklat pakai Naïve Bayes & XLSTAT. Klasifikasi produk akurat 93%, identifikasi preferensi konsumen terhadap 'Produk A' yang ideal.

Abstract

Penelitian ini mengeksplorasi penggunaan kombinasi metode CATA, XLSTAT, dan Naïve Bayes untuk mengklasifikasikan produk coklat berdasarkan atribut sensori dominan serta memahami preferensi konsumen. Check-All-That-Apply (CATA) merupakan metode evaluasi sensori deskriptif sederhana dan cepat untuk mengidentifikasi karakteristik produk berdasarkan persepsi konsumen walaupun bukan berasal dari panelis terlatih. Analisis klasifikasi menggunakan Naïve Bayes pada data CATA menghasilkan nilai rata-rata akurasi sebesar 93%. Begitu juga dengan precision, recall, dan F1-score menunjukkan nilai rata-rata diatas 90%. Prediksi terhadap 114 data uji menunjukkan 98 data sebagai ‘Produk A’, 16 data sebagai ‘Produk B’, dan tidak ada yang diprediksi sebagai ‘Produk C’. Hasil analisis XLSTAT mendukung temuan ini, menunjukkan bahwa produk ideal lebih terkait dengan Produk A, sebagaimana terlihat dalam grafik biplot yang menempatkan keduanya dalam kuadran F1 kanan, mengindikasikan kesamaan karakteristik yang dominan antara atribut produk ideal dengan Produk A.

Review

This study presents an intriguing integrated approach for classifying chocolate products and discerning consumer preferences by combining Check-All-That-Apply (CATA) methodology with Naïve Bayes algorithms and XLSTAT. The research aims to identify dominant sensory attributes through CATA, which is highlighted as a swift and straightforward descriptive sensory evaluation method using consumer perceptions rather than trained panelists. The subsequent analytical framework is designed to categorize products and link these classifications to consumer ideals, offering a valuable toolkit for product development and market understanding. A significant strength of this work lies in the robust performance of the Naïve Bayes classification applied to the CATA data. The reported average accuracy of 93%, alongside precision, recall, and F1-scores all exceeding 90%, indicates a highly effective model for distinguishing between product types based on sensory profiles. The classification of 114 test data points, predominantly identifying 'Produk A' (98 samples) and 'Produk B' (16 samples) with no prediction for 'Produk C', provides clear and actionable insights. Furthermore, the corroborative analysis using XLSTAT, which positions 'Produk A' and the ideal product within the F1 right quadrant of a biplot, strongly supports the notion that 'Produk A' embodies the desired sensory characteristics for consumers. While the abstract demonstrates a clear and effective methodology, a full manuscript could further enrich the discussion by delving into specific details. It would be beneficial to elaborate on which specific sensory attributes, as captured by CATA, are most influential in driving the high classification accuracy of the Naïve Bayes model. Additionally, exploring the implications of 'Produk C' not being predicted could offer insights into its unique sensory profile or market positioning. A discussion on the practical applications for product optimization and targeted marketing based on the identified dominant characteristics for 'Produk A' would also enhance the paper's impact and provide concrete takeaways for industry stakeholders.

Full Text

Comments

You need to be logged in to post a comment.