Testing data analysis system for the technology of harmful substances forecasting

Abstract

The issue of product safety is acute in many countries and companies. Indeed, every year a large number of people catch various diseases after using products contaminated with harmful substances. To increase food safety, there should be implemented some technologies which would enable quick testing of products and, using the obtained knowledge, to prevent appearance of harmful substances. Biosensor technologies which are utilized for conducting quick tests for agricultural purposes are being nowadays actively developed in the world. The use of data mining technologies can provide the identification of new knowledge about the nature and dependence of various harmful substances on external factors. The article considers the issues of construction and testing of forecasting technology, including analysis of existing solutions, search for forecasting models, usage of Data Mining methods to find qualitatively new knowledge and hypotheses, and hypothesis testing as well. The first section of the article provides an overview of existing models and methods, formulas for calculating the risk of aflatoxin. The second section contains a brief description of the data collection system which consists of a data warehouse, sensors and a mobile application as a data source. The results of data mining obtained after the use of Data Mining methods are presented in the third section of the paper. As a result, there was obtained a hypothesis of the effect on the appearance of mycotoxins in crops. Confirmation or refutation of the hypothesis was performed by testing it using OLAP-technologies. The results of the hypothesis test are presented in the last section. The main result of the work was obtaining a data analysis system ready for testing in real conditions. However, it should be noted that there is a need for more extensive testing on large amounts of data which can be accumulated over several years of observations.