An Examining from Data Sourced Deviations in Modelling by use of Variance Analysis Method

Ahmet KAYA
1.563 268

Abstract


In this study, outliers which can cause bias on models in a survey stage using observation  values  which  constitute  basement  for  conducting  modelling  have  been

investigated. The processes important for estimation observation,  transform to data, modelling from data, and to gain information from models so important by oneself.  One of the most important risks we encountered is transformation on data by naturel randomness or derivation by people. If transformation on data is derived by naturel randomness, solution is easy and has to compulsory ultimately. Causes for bias are appeared by time series (ARIMA-Auto Regressive Integrated Moving Average) models, detection, and detection causes on them are modelled by multi factored experimental design with and significance levels have been investigated.


Keywords


Time Series, Estimation, Modelling, Outliers, Variance Analysis.

Full Text:

PDF (Türkçe)

References


Bayhan, M., Kalite Kontrolünde Zaman Serisi Analizi, Endüstri Mühendisliği Dergisi, 21, 17-21, 1992.

Box, G. E. P., Jenkins G. M., Time series analysis: Forecasting and control, Sect 6.4.3. San Francisco, Holden-Day, 1976.

Chang, I., Tiao G. C., Chen, C., Estimation of Time Series Parameters in the Presence of Outliers, American Statistical Association and the American Society for Quality Control, 1988.

Fox, A. J., Outliers in Time Series, J. Royal Statistical Society B., 34(3), 350-363, 1972.

Kaya, A., AR(1) Modelinde A Tipi Sapan Etki, İstatistikçiler Dergisi, 3, 1-7, 2010.

Ljung, G. M., Box, G. E. P., The Likelihood Function of Stationary Autoregressive-Moving Average Models, Biometrika, 66, 265-270, 1979.

Ljung, G. M., On Outlier Detection in Time Series, J. Royal Statistical Society B, 55, 559-567, 1993.

Muirhead, C. R., Distinguishing Outlier Types in Time Series, J. Royal Statistical Society B, 48(1), 39-47, 1986.

Kaya, A., Outlier Effects On Databases, ADVIS 2004-Advances in Information Systems, Dokuz Eylül Üniversitesi, İzmir, 2004.

Kurt, S., Çok Etkenli Deneylerde Tek Sapan Değer Çözümlemesi, Seminer Çalışması, Ege Üniversitesi Fen Fakültesi İstatistik Bölümü, İzmir.