THE USE OF QUANTILE AND LINEAR REGRESSION MODELS IN PREDICTING MYOPIA

Ergül B., Altın Yavuz A.

3. BİLSEL INTERNATIONAL KIBYRA SCIENTIFIC RESEARCHES CONGRESS 19-20 OCTOBER, 2024/ BURDUR/ TÜRKİYE, Burdur, Türkiye, 19 - 20 Ekim 2024, cilt.1, sa.1, ss.338-343

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Cilt numarası: 1
  • Basıldığı Şehir: Burdur
  • Basıldığı Ülke: Türkiye
  • Sayfa Sayıları: ss.338-343
  • Eskişehir Osmangazi Üniversitesi Adresli: Evet

Özet

Myopia is a refractive error in the eye, commonly known as the inability to see distant objects clearly. Individuals
with myopia can see nearby objects clearly while distant objects appear blurry. This condition arises from
abnormalities in the structure of the eye, typically due to the eye being longer than normal or the cornea being
overly steep. As a result, distant objects appear blurry. There is a significant relationship between myopia and age,
with age being one of the factors that influence the onset, progression, and worsening of myopia.
Quantile regression is a method developed as an alternative to classical linear regression analysis. While classical
regression focuses on examining relationships between variables based on expected values, quantile regression
analyzes predictions for specific quantiles. In quantile regression analysis, separate optimizations are performed
for each quantile, aiming to minimize the predicted error term falling below a certain quantile. This method is
particularly effective in cases of heterogeneous data structures, where outliers are present or where variables exert
different effects in various segments of the data.
In this study, two significant variables related to myopia—axial length and vitreous chamber depth—were taken
as dependent variables separately, while age was treated as an independent variable. Regression analyses and
quantile regression analyses for various quantile values were conducted to compare which model provided better
predictions, leading to discussions about their effectiveness.