Chickpea yield and species prediction with machine learning

Increase in production in the agricultural sector; The expansion of cultivated agricultural areas depends on parameters such as timely maintenance, irrigation with appropriate methods and selection of appropriate seed types. Agricultural areas are insufficient to produce healthy and sufficient food for the increasing human population. It has been understood that in order to meet the need for healthy and sufficient food, it is necessary to obtain healthy and many products from the unit area. It has been seen that the prerequisite for increasing productivity is the selection of suitable seed species. This study was carried out in two stages. In the first stage, chickpea species were estimated by classifying the characteristics of the chickpea plant with machine learning methods, and in the second stage, the yield values of the species were estimated by regression method. In the separation of the data for testing and training, cross validation was 5 and 50% training was 50% testing. Decision trees (CA), support vector machines (SVM) and k nearest neighbor (KNN) methods from machine learning classification methods were found suitable and used. Accuracy was used as a measure of success. Three classes of raw data set, raw + synthetic data set and only synthetic data set were processed. When the success rates of the results obtained were compared, the highest success rate in the classification using only raw data was SVM with an accuracy rate of 90.6%. Likewise, the classification success rate of the dataset, in which synthetic data created together with the raw data, was used, was KA with the highest 100% accuracy rate. When the dataset consisting of only synthetic data was used, the highest success rate in classification was KNN with an accuracy rate of 95.4%. The high success performances have shown that machine learning methods can be used in this field.