Optimization of CoKriging Spatial Prediction Based on Ant Colony Algorithm and Genetic Algorithm

The purpose of this paper is to optimize the CoKriging spatial prediction based on ant colony algorithm and genetic algorithm to improve the accuracy of spatial estimation. Firstly, to address the sample scarcity situation, the data are preprocessed and converted to <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$266\times 266$</tex> square array, the semi-variance function is constructed after visualization, and the Kriging model is used to interpolate the prediction of the unsampled locations, to explore the relationship between the sample size and the mean squared error, and at the same time, the ant colony algorithm is introduced to optimize the Kriging model. Next, the correlation between the collaborative variables and the target variables is analyzed, the Pearson correlation coefficient and the Spearman rank correlation coefficient are calculated, the collaborative variables are selected, the collaborative variational function is constructed, the CoKriging model is used to estimate the spatial values of the unsampled locations, the sample size is adjusted for error analysis, and the genetic algorithm is used for optimization. Finally, the XGBoost algorithm was used to predict the unsampled location space values, and the average absolute error, mean square error and decidability coefficient were compared with the CoKriging method. The study shows that the ant colony algorithm and the genetic algorithm are effective in optimizing the Kriging model and the CoKriging model, respectively, and effectively reduce the mean square error, which provides a more effective method for spatial estimation.