Early cherry fruit pathogen disease detection based on data mining prediction

Milos Ilic, Sinisa Ilic, Srdjan Jovic, Stefan Panic

Research output: Contribution to journalArticle

5 Citations (Scopus)

Abstract

Today's world depends largely on information and communication technologies. These technologies are in use in different areas of human life and work. Each day, more examples of possible application of information and communications technology are being discovered. In most cases, computer science is used to solve complex problems which have mathematical background. The most important and challenging job in agriculture is plant protection. This is due to its complexity and the lack of specialized tools that could predict when the conditions for specific infections are fulfilled. In this paper authors use different mathematics-based techniques for data processing and prediction of possible fruit disease infection. Six significant weather variables and one variable representing the month in the year are selected as predictor variables. Implemented techniques are compared with each other in order to select the best. Prediction includes two most important diseases of cherry fruit: monilinia laxa and coccomyces hiemalis. Data sets used in this research include data of eight year time period, collected over the region of Toplica in Serbia. The best achieved prediction accuracy is 95.8%. Additionally, the same implemented methods can be applied on other fruit species and other diseases for which data are known.

Original languageEnglish
Pages (from-to)418-425
Number of pages8
JournalComputers and Electronics in Agriculture
Volume150
DOIs
Publication statusPublished - 1 Jul 2018
Externally publishedYes

Keywords

  • Classification tree
  • Data mining
  • Discriminant analysis
  • Linear regression
  • MATLAB

ASJC Scopus subject areas

  • Forestry
  • Agronomy and Crop Science
  • Computer Science Applications
  • Horticulture

Fingerprint Dive into the research topics of 'Early cherry fruit pathogen disease detection based on data mining prediction'. Together they form a unique fingerprint.

  • Cite this