Prediction of the Quality of Fresh Water in a Basin

Classification

Classification is one of the most used techniques of the multivariate static tool-box. It may be described roughly as a technique concerned with assigning data cases (i.e. observations) to one of a fixed number of possible classes, see Sharma (1996). Classifing is present when dealing with the needs grouping. It has been developed firstly by statisticians. Machine learning researchers have given new insights both from the theoretical and computation issues of classification. They consider that classification is a process that permits learning the existence of patterns. It ends fixing how cases must be classified different predetermined classes of data. See discussion on this problem in Brooks (2010), Hastie-Tibshirani-Friedman (2001). Once the model is built, it can be used to classify new data. Hence, the goal of Classification is sorting observations into two or more labeled classes. The emphasis of it is on deriving a rule that can be used to optimally assign new objects to the classes.

When classifying we assign a membership to each data item (case). Each record is related to a variable of interest , which allows classifying, and a set of covariates . Statisticians denominate the covariables “independent variables” and computer scientists “input variables”. A functional relationship between and should be described. Take a case (observation) . We evaluate, where , is the vector of attributes measured in case k and represents a vector of parameters which weights the importance of each coordinate of .