Geri Dön

Çok katlı regresyon analizinde kürecikler yöntemi ve bir uygulama

Başlık çevirisi mevcut değil.

  1. Tez No: 16414
  2. Yazar: İBRAHİM GÜNEY
  3. Danışmanlar: Belirtilmemiş.
  4. Tez Türü: Doktora
  5. Konular: İstatistik, Statistics
  6. Anahtar Kelimeler: Belirtilmemiş.
  7. Yıl: 1991
  8. Dil: Türkçe
  9. Üniversite: Uludağ Üniversitesi
  10. Enstitü: Fen Bilimleri Enstitüsü
  11. Ana Bilim Dalı: Belirtilmemiş.
  12. Bilim Dalı: Belirtilmemiş.
  13. Sayfa Sayısı: 89

Özet

Özet yok.

Özet (Çeviri)

ABSTRACT Multiple regression analysis is one of the statistical approach to prediction. In this analysis, one or more independent variables or predictors are selected along with one dependent variable. Using scors on these variables, it is possible to determine the least squares regression weights that minimize the squared distance between the predicted and actual values on the dependent variable.lt is also possible to predict with a certain degree of accuracy by using regression equation derived from the sample. If the interest of researchers lies in individual predictors standardized regression coefficient Cbeta weightsDis used to measure the importance of a predictors. These coefficients reflect the strength and direction of the relationship between the predictor variables and dependent variable. However,if the predictors are highly correlated,it is very difficult to determine the relative influence of each predictors on the dependent variable. This situation is called“Multicollineari ty”. When two or more predictors are highly correlated »they are measuring the same thing since much of the variance of one predictor is tibeing shared by the other. When one or more predictors are linear combinations of the remaining predictors, exact mul ti col 1 i near i ty exists »thus it is impossible to derive the unique estimates of the regression coefficients. The major effects of mul ti col linearity are described as follows: ID High variance of coefficients may drastically reduce the precision of estimation. 2D Esti mates of coefficients may be sensitive to particular sets of sample data. 3D Mul ti col linearity can result in coefficients appearing to have the wrong sign or opposite to the prior expectations of the researchers. 4DSome variables may be dropped from the model because they are not significant in the sample even thought they are important in the populations. Ridge Regression is a technique of developing the estimators of regression coefficients to remedy the mul ti col linearity problem. Ridge estimators are obtained by adding a constant into the normal equations and forcing the resulting estimators to be biased. When an estimators has only a small bias, but it is substantially more precise than an unbiased estimator, it may be preferred since it will have a a greater probability of being close to the true parameter value. A measure of the combined effects of bias and sampling illvariation is th© expected value of the squared deviation of the biased estimator from the true parameter ft. This measure is called“mean squared error”of the estimator and it can be shown that EC/XJO -ft! =VC^Ck:0 + CEC/3C10 -ftj* Mor ever, the mean squared error equals the variance of estimator plus the squared bias. If the estimator is unbiased, the mean squared error CMSED is identical to the variance of the esti mat or. The biased estimator can give a smaller mean squared error than the ordinary least squares COLSD estimator if the squared bias is no longer than the reduction achived in variance. This is the main aim of ridge regression. There were a few main purposes of this study. The first purpose was to demonstrate how ridge regression can be applied to data studied. The secondary purpose was to compare the char act eric tics and performance of the ridge methods versus the OLS methods. Al so, other purpose is to use the principle component analysis together the ridge regression to study the sample data. 3

Benzer Tezler

  1. Tez No

    857659

    Türkiye elektrik tüketiminin makine öğrenmesi yöntemleri ile tahmini

    Prediction of electricity consumption in turkiye with machine learning methods

    DENİZ HERSEK

    Yüksek Lisans

    Türkçe

    Türkçe

    2024

    İstatistikYıldız Teknik Üniversitesi

    İstatistik Ana Bilim Dalı

    PROF. DR. GÜLHAYAT GÖLBAŞI ŞİMŞEK

  2. Tez No

    712187

    Kentiçi demiryolunda trafik yükünün ray aşınmaları üzerindeki etkisinin araştırılması

    Investigation of the effects of traffic load on rail wear in urban railways

    HAZAL YILMAZ SÖNMEZ

    Doktora

    Türkçe

    Türkçe

    2021

    Ulaşımİstanbul Teknik Üniversitesi

    İnşaat Mühendisliği Ana Bilim Dalı

    PROF. DR. ZÜBEYDE ÖZTÜRK

  3. Tez No

    790421

    Dengesiz bir diyabet veri setinde makine öğrenmesi yöntemlerini kullanarak diyabet hastalığının teşhisi

    Diagnosis of diabetes disease using machine learning methods in an imbalanced diabetes dataset

    İSMAİL BUĞRA BÖLÜKBAŞI

    Yüksek Lisans

    Türkçe

    Türkçe

    2023

    Endüstri ve Endüstri MühendisliğiBursa Uludağ Üniversitesi

    Endüstri Mühendisliği Ana Bilim Dalı

    PROF. DR. BETÜL YAĞMAHAN

  4. Tez No

    66391

    Road surface micro and macro texture influence on skid resistance

    Yol yüzeyi mikro ve makrodokusunun kayma direncine etkisi

    MURAT ERGÜN

    Doktora

    İngilizce

    İngilizce

    1997

    İnşaat Mühendisliğiİstanbul Teknik Üniversitesi

    İnşaat Mühendisliği Ana Bilim Dalı

    PROF. DR. EMİNE AĞAR

  5. Tez No

    571851

    Psikoz şiddetini öngörmede kendini eleştirme ve diğer psikopatolojik süreçlerin etkilerinin sınanması

    Investigating the roles of self-criticism and psychopathological processes in predicting the severity of psychosis

    ZÜLAL ÇELİK

    Tıpta Uzmanlık

    Türkçe

    Türkçe

    2019

    PsikiyatriSağlık Bilimleri Üniversitesi

    Psikiyatri Ana Bilim Dalı

    DOÇ. DR. KAASIM FATİH YAVUZ

Geri Dön