The IRMA Community
Newsletters
Research IRM
Click a keyword to search titles using our InfoSci-OnDemand powered search:
|
Machine Learning in Studying the Organism’s Functional State of Clinically Healthy Individuals Depending on Their Immune Reactivity
Abstract
The work is devoted to the decision of two interconnected key problems of Data Mining: discretization of numerical attributes, and inferring pattern recognition rules (decision rules) from training set of examples with the use of machine learning methods. The method of discretization is based on a learning procedure of extracting attribute values’ intervals the bounds of which are chosen in such a manner that the distributions of attribute’s values inside of these intervals should differ in the most possible degree for two classes of samples given by an expert. The number of intervals is defined to be not more than 3. The application of interval data analysis allowed more fully than by traditional statistical methods of comparing distributions of data sets to describe the functional state of persons in healthy condition depending on the absence or presence in their life of the episodes of secondary deficiency of their immunity system. The interval data analysis gives the possibility (1) to make the procedure of discretization to be clear and controlled by an expert, (2) to evaluate the information gain index of attributes with respect to the distinguishing of given classes of persons before any machine learning procedure (3) to decrease crucially the machine learning computational complexity.
Related Content
|
.
© 2023.
34 pages.
|
|
.
© 2023.
15 pages.
|
|
.
© 2023.
15 pages.
|
|
.
© 2023.
18 pages.
|
|
.
© 2023.
24 pages.
|
|
.
© 2023.
32 pages.
|
|
.
© 2023.
21 pages.
|
|
|