GEE: Gradient Boosting Tree classification tutorial (sample production, feature addition, training, accuracy, parameter optimization, contribution, statistical area)

Author:CSDN @ _养乐多_

This article will introduce the method and code for Gradient Boosting Tree classification on the Google Earth Engine (GEE) platform, including tutorials on making sample points (mixing local, online and local online sample points, merging sample points, etc.) , add feature variables (various indices, texture features, time series features, phenological features, etc.), run the gradient boosting tree classifier tutorial, and apply the classifier model to pixel scale or superpixel (object/patch) scale data , calculate the accuracy of the gradient boosting tree classification results (the accuracy parameters are downloaded to the local in csv format), optimize the parameters of the gradient boosting tree classification algorithm (draw the optimal parameter distribution chart), and print the contribution of each variable feature (sorting feature contribution, Methods and codes for steps such as drawing histograms) and counting the area of ​​each land type.

This tutorial can be applied to a variety of classification scenarios, including land use/cover classification, planting area extraction (garlic, wheat, corn, etc.), local climate zone classification, vegetation classification, forest/grassland classification, disease/pest classification, flood prediction, etc. Various scenarios.

Gradient Boosting Tree classification process and classification results are shown in the figure below
————————————————————— ————————————————————————————

---