Crop Yield Prediction Using Deep Learning in Mediterranean Region

Abstract

Knowledge of meteorological or climatic data in a region is essential for the successful development of agriculture, energy and sustainable development in this region. The main goal of this article is the proper use of the data mining technique for meteorological and agricultural data to help in the development of agriculture in Mediterranean region. study of meteorological data affected by climate change using a data mining technique such as clustering technique by combining with knowledge base constructed from climate rules adapted to a specific agricultural crop. Using this technique, we can acquire new information that can help predict the future quality of the yield of this crop and sought to improve its production, the model built from the large dataset transfers the information retrieved in usable knowledge for classification and forecasting of climatic conditions. We discussed the use of a data mining technique to analyze meteorological and agricultural data. Various data extraction tools and techniques are already available, but they have been used in a very limited way for meteorological data and are never combined with a knowledge dataset adapted to a specific agriculture culture. In this paper, an algorithm based on a network of neurons to predict the impact of climate change on the production and yields of some agricultural crops for a future time and a given site.

The performance of our proposed ANN-based method was tested on a 30-year meteorological dataset comprising 54,000 records containing attributes such as temperature, humidity, wind velocity and rainfall as well as several agro-climatic data derived from the climate rules (Köppen classification). The prediction error turns out to be very low and the learning converges very strongly. Consequently, this article based on predictive data mining, will explore the possibility to extract interesting patterns or knowledge from a huge amount of meteorological and agro-climatic data.