Statistical prediction of grape production based on climatological and agronomic variables in Gramona S.A.

Identifier:  TFM:614

Handle: https://hdl.handle.net/20.500.11797/TFM614

Authors:  Sagrera Cots, Anna

Abstract:
The agricultural activity is dependent and is interconnected with the weather and the meteorology. Climate change and global warming are an unpredictable reality that especially affect agricultural crops and generate uncertainty and concern among farmers. The vine is one of the most climatologically sensitive crops characterized by large annual variations in quantity, quality and price in response to changes in weather conditions. This project studies and analyses a statistical yield prediction model based on climatological and agronomic variables, and a historical plantation production of Gramona S.A. The study is divided in two main parts. The first one compiles the agronomic and meteorological information of the area in order to analyse its evolution, generate a prediction of the present year with the Holt-Winters method and determine its importance and effects in the production according to CatBoost model. The most important variables are the age, annual temperature, rain from bud to harvest and harvest rain. In addition, those that have a positive effect on yield are high values of harvest rain, annual rain and minimum temperatures between March and April. The second part is based on the previous analysis. A model has been generated using the statistical program Unscrambler X version 10.2 to obtain performance values for the 2019 vintage. Moreover, the model has been evaluated to find out whether it is appropriate and could be extrapolated. Finally, the relative deviation of less than 20% indicates the correctness of the model considering that it has being based on highly unpredictable variables. This result can be used as a new standpoint of the sector and a prelude of its future.