Application of PLS algorithm in discriminant analysis in multidimensional data mining

Data mining technology emerges as the times require. Through the integrated learning of data, the original data is transformed into a form suitable for operation, useful data is extracted for mining, and finally, various strategies of data mining are applied to generate useful patterns and rules. Multiplicative regression has the potential of small computation, stable algorithm, easy to understand results, and maximum potential for mining data, which can be widely used in small-sample data mining. Aiming at the high-dimensional and low-sample problem, this paper constructs a small-sample mining algorithm using partial least squares (PLS) model and realizes dimension reduction and classification learning under the unified framework of PLS and in the classification of gene expression spectrum (colon) cancer data. To realize the mining and visualization of small-sample data by PLS. Compared with the classical algorithm SVMs, the results verify the validity and reliability of the PLS algorithm for high-dimensional and low-sample data mining problems.