Model strojnog učenja za rano predviđanje prinosa u pametnoj poljoprivredi

master's thesis

• Title (english): Machine learning model for early yield prediction in smart agriculture
• Author: Fran Kozjak
• Mentor: Gordan Ježić (mentor)
• Committee member: Gordan Ježić (predsjednik povjerenstva)
• Committee member: Ivana Podnar Žarko (član povjerenstva)
• Committee member: Mario Kušek (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2023-07-04, Croatia

Abstract

Pametna poljoprivreda je sve više u fokusu agronomske zajednice i industrije zbog svoje sposobnosti poboljšanja učinkovitosti proizvodnje i smanjenja troškova. Jedan od ključnih čimbenika u postizanju ovih ciljeva je upotreba tehnologija poput strojnog učenja za predviđanje prinosa usjeva. Predviđanje prinosa je ključno za donošenje važnih odluka u poljoprivrednom sektoru, kao što su planiranje proizvodnje, upravljanje zalihama i tržišnim cijenama. U ovom radu predstavljeni su modeli strojnog učenja za rano predviđanje prinosa u pametnoj poljoprivredi te se određuju najprecizniji od njih. Cilj ovog istraživanja je pružiti efikasno rješenje za predviđanje prinosa usjeva na temelju korištenja tehnologije strojnog učenja.Rad opisuje načine rada modela, njihovu arhitekturu, kao i korake u pripremi podataka za treniranje modela. Ovaj rad pruža korisne smjernice za razvoj sličnih modela u budućnosti, kao i poboljšanje proizvodnje u poljoprivrednom sektoru.

Abstract (english)

Smart agriculture is increasingly the focus of the agronomic community and industry for its ability to improve production efficiency and reduce costs. One of the key factors in achieving these goals is the use of technologies such as machine learning to predict crop yields. Yield forecasting is essential for making important decisions in the agricultural sector, such as production planning, inventory management and market prices. In this paper, machine learning models for early yield prediction in smart agriculture are presented and the most accurate of them are determined. The goal of this research is to provide an efficient solution for predicting crop yields based on the use of machine learning technology. This paper describes the modes of operation of the models, their architecture, as well as the steps in the preparation of data for model training. This paper provides useful guidelines for the development of similar models in the future, as well as improving production in the agricultural sector.

• Keywords: Pametna poljoprivreda; Kukuruz; Strojno učenje; Linearna regresija; SVM; Stablo odlučivanja; Random forest; Neuronska mreža; Prenaučenost; predviđanje prinosa
• Keywords (english): Smart agriculture; Maize; Machine learning; Linear regression; SVM; Decision tree; Random forest; Neural network; overfitting; yield prediction
• Language: croatian
• URN:NBN: urn:nbn:hr:168:023515
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-10-11 09:23:00