Pernicious bird density distribution prediction method based on multi-model fusion Stacked

The invention discloses a pernicious bird density distribution prediction method based on multi-model fusion Stacked. The pernicious bird density distribution prediction method comprises the steps of step A, selecting a feature-rich region; step B, carrying out data acquisition based on the selected rich area as a sample, carrying out feature processing before model training on the acquired data, and carrying out grading processing according to different ranges of density of pernicious bird species; step C, building and training a first layer Stacked model; step D, building and training a secondary-layer Stacked model; step E, in learning training of the first-layer Stacked model and the second-layer Stacked model, carrying out parameter tuning based on a grid search method, and selecting the first-layer Stacked model and the second-layer Stacked model containing optimal parameters; and step F, combining and optimizing the first-layer Stacked model and the second-layer Stacked model to obtain a prediction mod