Bibliographic citations
Mariscal, V., (2016). Integración de datos y análisis predictivo en tratamiento de drogodependencia [Trabajo de investigación, Universidad Complutense de Madrid]. http://renati.sunedu.gob.pe/handle/sunedu/952182https://eprints.ucm.es/39325/
Mariscal, V., Integración de datos y análisis predictivo en tratamiento de drogodependencia [Trabajo de investigación]. ES: Universidad Complutense de Madrid; 2016. http://renati.sunedu.gob.pe/handle/sunedu/952182https://eprints.ucm.es/39325/
@mastersthesis{renati/1245,
title = "Integración de datos y análisis predictivo en tratamiento de drogodependencia",
author = "Mariscal Carhuamaca, Victor Hugo",
publisher = "Universidad Complutense de Madrid",
year = "2016"
}
The current data analysis faces problems arising from the combination of data from various sources. The value of information can be enhanced greatly facilitating the integration of new data sources and industry is well aware of it today. However, not only the volume but also the great diversity of data is a problem prior to analysis. A good integration of data ensures reliable results and therefore worth stopping in the specification process improvement, collecting, cleaning and data integration. This work is dedicated to the cleaning phase and data integration analyzing existing procedures and proposing a solution that applies to medical data, thus focusing on projects prediction (with purpose of prevention) in health sciences. In addition to the implementation of cleaning processes, develop algorithms of detection of outliers that allow improving the quality of the data set after being eliminated. The work also includes the implementation of a process of prediction that serve as an aid to decision-making. Specifically this work performs a predictive analysis of the data of patients drug addicts of the clinic Nuestra Señora de la Paz, in order to be able to offer support in decisions of the physician in charge admit the internment of patients in the clinic In the majority of cases the study of data provided requires a proper pre-procesado to traditional statistical analysis results to be reliable. So in this paper are implemented various ways to detect the outliers: an own algorithm (Detection of Outliers not monotonous chains), that uses the advantages of the algorithm Knuth-Morris-Pratt for pattern recognition, and the bookshops outliers and Rcmdr of R. The application of cleaning procedures and data integration and elimination of outliers provides a clean and reliable base data on which prediction procedures be implemented data with Naive Bayes classification algorithm in R.
File | Description | Size | Format | |
---|---|---|---|---|
Mariscal_Carhuamaca_VH.pdf Restricted Access | TFM (abierto en repositorio de origen) | 5.14 MB | Adobe PDF | View/Open Request a copy |
Autorizacion.pdf Restricted Access | Formato de autorización | 793.2 kB | Adobe PDF | View/Open Request a copy |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.