Abstract
When using data mining or machine learning techniques on large and diverse datasets, it is often necessary to construct descriptive and predictive models. Descriptive models are used for discovering relationships among the attributes of the data while predictive models identify the characteristics of the data that will be collected in future. Bioinformatics data are high-dimensional, making it practically impossible to apply the majority of "classic" algorithms for classification and clustering. Even when the algorithms are useful, the training with large multidimensional data significantly increases the processing time. The algorithms specialized for working with high-dimensional data often cannot process data that contains large data sets that have several thousand dimensions (features). Dimension reduction methods (such as PCA) do not provide satisfactory results, and in addition, they obscure the meaning of the initial attributes in the data. For the constructed models to be usable, they must meet the requirement of scalability due to the large increase in the amount of bioinformatics data collected daily. Furthemore, the significance of the individual data features can also differ from source to source. This work describes an attribute selection method to efficiently classify high-dimensional (30,698) transcriptomics data collected from multiple sources. The proposed method was tested using 22 classification algorithms. The classification results for the selected sets of attributes are comparable to the results for the complete set of attributes.