Predictor Envelopes and Standard Regression Models: An Empirical Juxtaposition
Journal: International Journal of Science and Research (IJSR) (Vol.8, No. 1)Publication Date: 2019-01-05
Authors : Joseph Bamidele Odeyemi;
Page : 1233-1237
Keywords : Renvlp software; standard regression; sufficient dimension; predictor envelopes;
Abstract
This study compares predictor envelopes and standard regression models using data on the attributes (Weight, Height, Shoe Size, Chest Diameter, Diastolic Blood Pressure, Fasting Blood Sugar, and Age) of pregnant women obtained from General Hospital, Offa. The purpose is to use both methods to fit models and determine which of the two is more efficient in prediction. The study applied predictor envelopes method as well as standard regression method to the data and was analyzed with the support of Renvlp statistical software. It was found that the information criterion AIC and BIC agreed to u=2 sufficient dimension. It was also discovered that both methods declare variable height as the only active predictor of weight of pregnant women due to their estimates (51.07, 52.03), z-scores (0.44, 0.45), and ratios of asymptotic standard error (0.99) The predictive performance of the two models show that the standard error of predictor envelopes estimates are smaller than those of the standard regression estimates. Thus, Predictor envelope is better than standard regression method even with one response variable y.
Other Latest Articles
- Chronic Subdural Haematoma: Aeromedical Disposition
- Comparison of Job Satisfaction in Occupational Therapy Settings
- Protocols for the Treatment of Stretch Marks of the Skin: A Systematic Review of Intervention
- Analysis, Resolution Differentiation of Principal Component Analysis and Correspondent Analysis
- A Study of Serum Magnesium Level in Patients with Type 2 Diabetes Mellitus in Relation to Its Micro Vascular and Macro Vascular Complications
Last modified: 2021-06-28 17:20:55