R. Moragues, J. Aparicio, M. Esteve

En esta presentación, introducimos un método basado en el aprendizaje no supervisado para la estimación de fronteras de producción. Este nuevo método, llamado uDEA, generaliza las técnicas clásicas de Análisis Envolvente de Datos (DEA), y satisface axiomas microeconómicos fundamentales como la convexidad y la libre disponibilidad de inputs y ouputs, pero a diferencia del DEA tradicional, no satisface el principio de mínima extrapolación. Este método está basado en el algoritmo OneClass Support Vector Machine para la estimación del soporte de una distribución estadística, con una función de transformación lineal a trozos, y tiene como objetivo reducir el problema de sobreajuste presente en el DEA. También introducimos la manera de medir la ineficiencia técnica. Evaluamos el método a través de una experiencia computacional, que muestra que el error cuadrático medio de la frontera estimada por el uDEA es hasta un 83% mejor que la estimada por el DEA estándar en ciertos escenarios.

Keywords: Análisis Envolvente de Datos, aprendizaje no supervisado, Máquinas de Soporte Vectorial, eficiencia técnica, sobreajuste

Scheduled

Data Envelopment Analysis II
June 10, 2022  4:00 PM
Conference hall


Other papers in the same session

Gradient Tree Boosting para la estimación de fronteras de producción

M. D. Guillén, J. Aparicio Baeza, M. Esteve Campello

Ranking de investigación de las universidades públicas españolas realizado con DEA

F. Fernández Palacín, M. A. López Sánchez, M. Muñoz Márquez

Efficiency measure in a fully fuzzy DEA

M. Arana Jiménez, A. Younesi, M. C. Sánchez Gil, S. Lozano Segura


Latest news

  • 6/8/22
    Visit Alhambra

    Tomorrow, Thursday, we will visit the Alhambra. 1. At 7:00 p.m., buses will start leaving from the door of the Hotel Granada Center (in front of the door of the Faculty of Sciences) 2. Upon arrival, groups with guides will be organized. 3. There will be groups in English language. These groups will wait at the destination until the arrival of all buses. 4. After the visit to the Alhambra we will walk to the Carmen de los Mártires to have a cocktail. 5. On the way back, we can go back by bus, which will leave from the same place where they left us at the Alhambra. 6. Although the way to the Alhambra by bus seems long, it really is close to the city. We invite you, whoever wishes, to go down for a pleasant walk. 7. Do not forget to take your Alhambra ticket and your identity card or passport. The document must be the same whose number matches the ticket number you have.

  • 5/31/22
    YouTube Channel

    The menu label is already on the web from where you can see the live broadcasts: inauguration, plenary sessions, round tables,... Don't forget to enter the YouTube channel and subscribe!!!!

  • 5/27/22
    Gala Dinner

    It opens the possibility that anyone who is registered for the conference, and wants non-registered companions to attend the gala dinner, can do so by paying the price of this dinner (80 euros VAT included). This payment must be made before June 1 (inclusive). To do this, contact with [email protected]

  • 5/23/22
    IMPORTANT WARNING

    To access the Alhambra on the visit on Thursday 9, it is obligatory to present the identification document whose reference was sent to the organization (eurocongress)

  • 5/16/22
    Social Events

    All social events are included with the payment of the corresponding registration.

Cookie policy

We use cookies in order to be able to identify and authenticate you on the website. They are necessary for the correct functioning of it, and therefore they can not be disabled. If you continue browsing the website, you are agreeing with their acceptance, as well as our Privacy Policy.

Additionally, we use Google Analytics in order to analyze the website traffic. They also use cookies and you can accept or refuse them with the buttons below.

You can read more details about our Cookie Policy and our Privacy Policy.