National and Kapodistrian University of Athens
Διακριτή l1 προσέγγιση υπό αύξουσες αποδόσεις και εφαρμογές σε οικονομικά δεδομένα
Abstract
dc:descriptionThe problem of convexity runs deeply in economic theory and practice. For example, increasing returns or upward slopes, which provide convexity, and diminishing returns or downward slopes, which provide concavity, of certain supply, demand, production and utility relations are often assumed in economics. Quite frequently, however, the observations have lost convexity or concavity due to errors of the measuring process, occasionally few of them being outliers. We address the question of smoothing the data by minimizing the sum of the moduli of the errors subject to the condition that the second divided differences of the smoothed data are nonnegative, that is, by assuming non-decreasing returns of the changed data. It follows that the best piecewise linear interpolant to the smoothed data is a convex curve, while in short, we address the problem of calculating a best L1 convex fit to the data. This is a linear inequality constrained optimization problem, which can be expressed as a linear programming calculation. In this dissertation we establish optimality conditions for the characterization of the best L1 convex fit, in terms, both of certain parameters that resemble the Karush-Kuhn-Tucker multipliers, when the objective function is replaced by a differentiable one, and of certain parameters that depend on the interpolation points and the KKT-like parameters. Also, we develop an efficient descent algorithm that is much faster than general linear programming procedures. Its efficiency is due to the banded matrices that occur during the iterations. Intensive numerical results illustrate the computation and present the efficacy of the method in small, medium and large data sets with errors that range from few outliers to a full corruption of the data. Certain applications are presented in order to illustrate the convexity assumption of real economic time series. The results are analyzed and the interpretation capability of the method is presented, due to its rich inherent structure, the values of the KKT-like parameters and the values of the parameters at the interpolation points.
Degree
thesis:*- Grantor dc:publisher
- National and Kapodistrian University of Athens
- Year dc:date
- 2012
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Papakonstantinou, Sotirios
- Παπακωνσταντίνου, Σωτήριος
Subjects
dc:subject × 30- L1 προσέγγιση
- Άθροισμα απολύτων τιμών σφαλμάτων
- Αύξηση πληθυσμού
- Αύξουσες αποδόσεις
- Γραμμικός προγραμματισμός
- Δείκτης ανθρώπινης ανάπτυξης
- Διηρημένη διαφορά
- Επιτόκιο
- Ζήτηση χρήματος
- Κατά κεφαλήν ΑΕΠ
- Brent
- Concavity
- Convexity
- Data fitting
- Diminishing returns
- Divided difference
- Gini coefficient
- GNI pc
- Human development index (HDI)
- Increasing returns
- Φυσικές Επιστήμες
- Μαθηματικά
- Επιστήμη Ηλεκτρονικών Υπολογιστών και Πληροφορική
- Κοινωνικές Επιστήμες
- Οικονομικά και Επιχειρήσεις
- Natural Sciences
- Mathematics
- Computer and Information Sciences
- Social Sciences
- Economics and Business
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/29262