Time series analysis of agro-meteorological through algorithms scalable data mining case: Chili river watershed, Arequipa

Abstract

The paper proposes a model for predicting climate change, using algorithms in mining techniques based on approximate data, applied to agro-meteorological data, by identifying groups search of motifs and time series forecasting. To achieve the goal you work with the water balance components: flow, precipitation and evaporation; also took into account the climatic variety seasons marked by humidity (December, January, February, March) and dry (other months) providing better to abstract sub-classification for temporary data processing three classification techniques: linear regression, Naive Bayes and neural networks, where the results of each algorithm are compared with other results. Then the mathematical method of linear regression predicting water balance components for a period of approximately 12 months on the data of dams Pane and Fraile Water Resources in River Basin Chili, Arequipa is performed.