Mathematical Optimization as A Tool for the Development of "Smart" Agriculture in Kazakhstan

článek v časopise ve Scopus, Jsc

en

This article uses methods for predicting plant performance indicators in Kazakhstan. In the work, using deep learning, visualization of predicted indicators (indicators and others), statistics from predicted values and identified changes, time series have been developed. Sentinel satellite data and statistical indicators for the last few years for the agricultural territories of Kazakhstan are used as primary data. It is found that the upward trend in wheat quality, however, increases the size of fertilizers, variables based on the NDVI also significantly contribute to the forecasting model. It has been shown that the amount of applied fertilizer has stabilized in the past few years due to economic and environmental constraints, so NDVI-based models will become increasingly important for enhancing forecasting models. Four machine learning algorithms have been evaluated and compared, namely boosted regression trees (BRT) and support vector machine (SVM), to map and predict the field yield of the Experimental Oil Farm in East Kazakhstan using readily available additional data. Based on the results of the work, a forecast of crop yields and general statistical recommendations for increasing yields were obtained. © 2021, AIDIC Servizi S.r.l.

This article uses methods for predicting plant performance indicators in Kazakhstan. In the work, using deep learning, visualization of predicted indicators (indicators and others), statistics from predicted values and identified changes, time series have been developed. Sentinel satellite data and statistical indicators for the last few years for the agricultural territories of Kazakhstan are used as primary data. It is found that the upward trend in wheat quality, however, increases the size of fertilizers, variables based on the NDVI also significantly contribute to the forecasting model. It has been shown that the amount of applied fertilizer has stabilized in the past few years due to economic and environmental constraints, so NDVI-based models will become increasingly important for enhancing forecasting models. Four machine learning algorithms have been evaluated and compared, namely boosted regression trees (BRT) and support vector machine (SVM), to map and predict the field yield of the Experimental Oil Farm in East Kazakhstan using readily available additional data. Based on the results of the work, a forecast of crop yields and general statistical recommendations for increasing yields were obtained. © 2021, AIDIC Servizi S.r.l.

mathematical; optimization; tool; development; smart; agriculture; kazakhstan

15.11.2021

Italian Association of Chemical Engineering - AIDIC

2283-9216

88

1219–1224

6