There are many methods of data mining that can work well on low-dimensional data, whereas data sets as raw materials in data mining generally have high dimensions so that in addition to the necessary initial processing such as aggregation and binary, dimensionality reduction is also important. If dimensionality reduction of data is done directly by removing certain features, there are characteristics of data to be lost, and it could be discarded feature is the most influential feature.
Dimensionality data reduction methods work in a certain way to capture data characteristics by mapping data sets from their original dimensions to other relatively low dimensions. This mapping produces a major component which can then be retrieved by components of a new dimension that have a major influence on the data set and dispose of components that have no major effect. The widely used methods are Principal Component Analysis and Singular Value Decomposition.
Principal Component Analysis mapping or transforming sets of data from the old dimension to a relatively new dimension with lower dimension using techniques in linear algebra, without requiring certain parameter inputs to provide the output of the mapping.
Analysis performed by Singular Value Decomposition has many similarities with PCA. SVD also needs to reduce any data on feature with average, although in fact SVD also allows not to reduce with an average for data that has been relatively rare. If PCA generally uses eigenvalue and eigenvectors to get a solution, SVD though will be same as PCA using single value decomposition to get a solution
Dimensionality data reduction methods work in a certain way to capture data characteristics by mapping data sets from their original dimensions to other relatively low dimensions. This mapping produces a major component which can then be retrieved by components of a new dimension that have a major influence on the data set and dispose of components that have no major effect. The widely used methods are Principal Component Analysis and Singular Value Decomposition.
Principal Component Analysis mapping or transforming sets of data from the old dimension to a relatively new dimension with lower dimension using techniques in linear algebra, without requiring certain parameter inputs to provide the output of the mapping.
Analysis performed by Singular Value Decomposition has many similarities with PCA. SVD also needs to reduce any data on feature with average, although in fact SVD also allows not to reduce with an average for data that has been relatively rare. If PCA generally uses eigenvalue and eigenvectors to get a solution, SVD though will be same as PCA using single value decomposition to get a solution
Advertisement
No comments