Back to search

Institutional Repository of Vilnius University

Prekybos tinklų prekių klasifikavimas naudojant kvitų duomenis /

Abstract

dc:description

The Consumer Price Index (CPI) is one of the most essential macroeconomic indicators used to measure price changes and assess the inflation rate. This indicator helps evaluate a country’s economic status and make the right decisions. Since the beginning of 2026, supermarket scanner data has been included in Lithuania’s CPI calculations. This data is characterized by a large amount of short texts such as product names, descriptions, or their abbreviations. In order to use this data for CPI calculations, every product must be assigned to a corresponding category of the Classification of Individual Consumption According to Purpose (COICOP) 2018 classifier, that is precisely why it is important to ensure proper data preparation for a more efficient use. There have been various types of data preprocessing in this work – correction of incorrect barcodes, text cleaning of unnecessary symbols or numbers, filling with relevant symbols or numbers, and data quality improvement by standardizing values according to best‐practice examples. A part of the missing values was filled by grouping goods according to their barcodes. TF–IDF and Word2Vec vectorization methods, as well as logistic regression, support vector machine and fastText classifiers, were applied. Hyperparameters yielding the best classification performance according to the macro‐F1 metric were selected. The best classification models of each classifier were compared with each other. Although the obtained results did not differ that much, while taking into account additional practical aspects, it has been established that the most suitable approach to the scanner data classification problem is the combination of TF–IDF vectorization with a Support Vector Machine (SVM) classifier. In addition, problematic subclasses that require further attention were identified.

Degree

thesis:*
Grantor dc:publisher
Institutional Repository of Vilnius University
Year dc:date
2026

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Noreikaitė, Ieva Marija,

Rights

dc:rights
Statement dc:rights
  • info:eu-repo/semantics/openAccess
Language dc:language
lit

Identifiers

dc:identifier.*

Chain of custody

source
Harvested from
Vilnius University
Base URL
epublications.vu.lt/oai
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
related terms
citation

Noreikaitė, Ieva Marija,. Prekybos tinklų prekių klasifikavimas naudojant kvitų duomenis /. Institutional Repository of Vilnius University, 2026. https://repository.vu.lt/VU:ELABAETD308334797&prefLang=en_US