Digital Internship: Principal Component Analysis (PCA) for Honey Fingerprinting and Quality Assessment

University of Split
5-10 weeks
5-10 hours per week

This digital internship, organised by the Faculty of Science, requiring approximately 50 hours of
student workload, focuses on applying Principal Component
Analysis (PCA) – a linear algebra technique based on matrix
diagonalisation of linearly correlated variables – to analyse the
characteristics of honey.

The project will use a transdisciplinary dataset in which various
honey samples are characterised by features or variables
relating to basic composition, colour, electrical conductivity,
viscosity, elasticity, and biomedical and environmental
characteristics.

Tasks and duties entrusted to the student:

The student's objective is to use PCA to reduce this
multivariable data to a few key Principal Components,
enabling fingerprinting of different honey samples and
identification of the underlying features (combinations of
variables) that best discriminate between high-quality,
authentic samples and those that are adulterated or
mislabelled by geographical origin.

Contact persons:

  • Assist. Prof. Dr. Marina Kranjac
  • Assist. Prof. Dr. Mija Marinković
  • Assist. Prof. Dr. Gordan Radobolja
  • Assist. Prof. Dr. Tomislav Rončević
  • Assoc. Prof. Dr. Jadranka Šepić
  • Prof. Dr. Mile Dželalija

Assessment method(s): Poster presentation or a report/article of the Internship
activities and results.

Skills to be acquired or developed:

Upon successful completion of the Digital Internship,
students will be able to demonstrate proficiency across
transdisciplinary integration, computational analysis,
mathematical application, and scientific communication,
with a specific learning outcomes:
• Articulate how the simultaneous analysis of physical,
chemical, and environmental properties collectively
determine the biomedical value and authenticity of
honey.
• Successfully implement, execute, and evaluate the PCA
algorithm on a high-dimensional dataset using standard
data science tools.
• Describe and explain the mathematical role of matrix
diagonalization (eigenvalue decomposition of the
covariance matrix) in transforming raw, correlated
variables into uncorrelated Principal Components.
• Interpret the loadings of the Principal Components to
identify which combinations of variables are the primary
drivers of variability between different honey samples.
• Effectively synthesize, communicate, and visualize
complex statistical findings to present a data-driven
protocol for honey quality screening.

Upon successful completion of the internship, the student will be awarded a
certificate. Recognition of the internship is subject to the university’s internal
policies.

Assist. Prof. Dr. Marina Kranjac, mkranjac@pmfst.hr