Comprehensive analysis of deep and machine learning approaches for predicting crop yields.

Crop selection decisions and measures taken during the crop's growing season are all supported by predicting crop output, which relies heavily on machine learning, a vital tool in the agricultural domain. Various ML techniques have been employed to support the prediction of agricultural production in research endeavours. The methods and characteristics used in research on crop yield prediction were extracted and synthesized as part of this investigation's comprehensive analysis. We found 567 relevant studies from six internet databases using our preset search parameters. Then, carefully chose 50 papers using specified inclusion and exclusion criteria for in-depth analysis. The Methodology and characteristics of these selected studies were extensively reviewed, providing recommendations for more studies. Based on the data, soil type (ST), temperature (TEMP), and rainfall (R) emerge as the predominant factors utilized in crop yield prediction models. Moreover, regarding these models, Artificial Neural Networks (ANNs) are the most commonly utilized algorithm. Then, to further our knowledge, we searched electronic databases to locate papers that used deep learning methods. It can extract the deep learning techniques from the 30 relevant publications produced by this search. This further research shows that CNN emerged as the dominating DL method in these investigations, along with the frequent application of DNN and LSTM. [ABSTRACT FROM AUTHOR]