Local cover image
Local cover image
Local cover image
Local cover image

A K-nearest neighbor imputation method based on locally weighted scatterplot smoothing (LOESS) for ordinal data

By: Material type: TextTextLanguage: English Publication details: 2008; 2008Description: 54 leavesSubject(s): Dissertation note: Thesis (BS Applied Mathematics) -- University of the Philippines Mindanao, 2008 Abstract: This study introduced a modification of the classical k-nearest neighbor algorithm called Weighted KNN based on Locally Weighted Seatterplot Smoothing (LOESS). It uses weighting scheme to make sure that the degree of influence of each neighbor is accounted in estimating missing values. The newly developed imputation technique was evaluated and compared to the existing methods using Dermatology and Breast Cancer data sets. The incomplete data sets in the experiment were generated under MCAR condition only with 1%, 5%, 10% and 15% levels of missing values. K-means and k-modes clustering algorithms were used to determine the recovery of each technique. Results showed that Weighted KNN based on LOESS performed best when applied on both Dermatology and Breast Cancer data sets with K-means clustering algorithm. It ranked next to KNN when applied on Dermatology data set with k-modes clustering algorithm and other outperformed the rest of the techniques on Breast Cancer data set. In general, Weighted KNN based on LOESS showed promising results when tested on both data sets.
Tags from this library: No tags from this library for this title. Log in to add tags.
Star ratings
    Average rating: 0.0 (0 votes)
Holdings
Cover image Item type Current library Collection Call number Status Date due Barcode
Thesis Thesis University Library General Reference Room-Use Only LG993.5 2008 A64 A66 (Browse shelf(Opens below)) Not For Loan 3UPML00012282
Thesis Thesis University Library Archives and Records Preservation Copy LG993.5 2008 A64 A66 (Browse shelf(Opens below)) Not For Loan 3UPML00032508

Thesis (BS Applied Mathematics) -- University of the Philippines Mindanao, 2008

This study introduced a modification of the classical k-nearest neighbor algorithm called Weighted KNN based on Locally Weighted Seatterplot Smoothing (LOESS). It uses weighting scheme to make sure that the degree of influence of each neighbor is accounted in estimating missing values. The newly developed imputation technique was evaluated and compared to the existing methods using Dermatology and Breast Cancer data sets. The incomplete data sets in the experiment were generated under MCAR condition only with 1%, 5%, 10% and 15% levels of missing values. K-means and k-modes clustering algorithms were used to determine the recovery of each technique. Results showed that Weighted KNN based on LOESS performed best when applied on both Dermatology and Breast Cancer data sets with K-means clustering algorithm. It ranked next to KNN when applied on Dermatology data set with k-modes clustering algorithm and other outperformed the rest of the techniques on Breast Cancer data set. In general, Weighted KNN based on LOESS showed promising results when tested on both data sets.

There are no comments on this title.

to post a comment.

Click on an image to view it in the image viewer

Local cover image Local cover image
 
University of the Philippines Mindanao
The University Library, UP Mindanao, Mintal, Tugbok District, Davao City, Philippines
Email: library.upmindanao@up.edu.ph
Contact: (082)295-7025
Copyright @ 2022 | All Rights Reserved