Please use this identifier to cite or link to this item:
Title: Anàlisi discriminant
Author: Cañas Porcuna, María José
Director: Florit i Selma, Carmen
Keywords: Anàlisi multivariable
Treballs de fi de grau
Teoria de la predicció
Anàlisi de regressió
Multivariate analysis
Bachelor's thesis
Prediction theory
Regression analysis
Issue Date: 17-Jan-2016
Abstract: It is common to find the necessity of identifying the characteristics that allow to distinguish two or more groups of individuals. The discriminant analysis consists in studying and analysing these characteristics that you can use at the time of classifying in two or more groups. To know how to distinguish the groups you need to get the information, evaluated in variables in how it is supposed to distinguish. With the discriminant analysis you can find these variables and which of these are necessary to achieve the best classification. You use the name of class to identify the groups as an answer, for example, a categoric variable with as many discreet values as groups they have. The variables that are used to distinguish the groups are used as predictors or discriminant variables. There are several ways to deal with these analysis. In particular, this work is focused on the classical discriminant analysis: Fisher’s linear discriminant and quadratic discriminant. Another technique is the canonical analysis of populations that it is applied in the case of more than two populations with the objective of representing them in orthogonal axes that allow to explain better the relation between the different groups. It will also be treated the logistic’s regression model and the discriminant based on distances.
Note: Treballs Finals de Grau de Matemàtiques, Facultat de Matemàtiques, Universitat de Barcelona, Any: 2016, Director: Carmen Florit i Selma
Appears in Collections:Treballs Finals de Grau (TFG) - Matemàtiques

Files in This Item:
File Description SizeFormat 
memoria.pdfMemòria567.9 kBAdobe PDFView/Open

This item is licensed under a Creative Commons License Creative Commons