قابلیت پیش بینی مدل‌های آماری در ارزیابی توان تولید رویشگاه راش شرقی
محورهای موضوعی : راشسمیه دهقانی نژاد 1 , سید جلیل علوی 2 , سید محسن حسینی 3
1 - فارغ التحصیل کارشناسی ارشد دانشگاه تربیت مدرس
2 - استادیار گروه جنگلداری، دانشکده منابع طبیعی و علوم دریایی، دانشگاه تربیت مدرس
3 - استاد گروه جنگلداری، دانشکده منابع طبیعی و علوم دریایی، دانشگاه تربیت مدرس، نور، ایران
کلید واژه: اعتبارسنجی, توان تولید رویشگاه, مجذور میانگین مربعات خطا, مدل خطی تعمیم یافته, مدل جمعی تعمیم یافته,
چکیده مقاله :
در مطالعه حاضر قابلیت پیش بینی مدل‌های خطی و جمعی تعمیم یافته برای ارتفاع غالب گونه راش به عنوان شاخصی عالی از کیفیت رویشگاه مورد بررسی قرار گرفته است. ارتفاع غالب در مطالعه حاضر به صورت میانگین ارتفاع سه اصله از مرتفع ترین درخت در هر قطعه نمونه تعریف می‌شود. به این منظور، در تیپ‌هایی که در آنها گونة راش غالب بود، 127 قطعة نمونه دایره‌ای به مساحت 1000 مترمربع پیاده و در هر یک از آنها ارتفاع و قطر تمام درختان گونة راش قطورتر از 5/7 سانتی‌‌متر علاوه بر ارتفاع از سطح دریا و درصد شیب و آزیموت اندازه‌گیری و ثبت شد. همچنین در مرکز هر قطعة نمونه، از عمق 10-0 سانتی‌متری، نمونه‌برداری خاک صورت گرفت و متغیرهای فیزیکی و شیمیایی متعددی اندازه‌گیری شد. در تحقیق حاضر، عملکرد پنج روش گزینش متغیر به‌طور جداگانه برای هریک از مدل‌های خطی و جمعی تعمیم‌یافته مورد بررسی قرار گرفته است. به منظور مقایسه کارآیی روش‌های گزینش متغیر در مدل خطی تعمیم‌یافته، از اعتبارسنجی متقابل با 2500 تکرار و در مدل جمعی تعمیم‌یافته از اعتبارسنجی fold-10 استفاده شده است. پس از انتخاب بهترین روش گزینش متغیر در هریک از دو مدل خطی و جمعی تعمیم‌یافته، اهمیت نسبی هر یک از متغیرهای مهم را محاسبه نموده که درنهایت متغیر ارتفاع از سطح دریا به‌عنوان مهمترین متغیر اثرگذار بر ارتفاع غالب گونه راش شناسایی شد. سپس با استفاده از معیارهای ارزیابی حاصل از داده‌های مدل‌سازی مشاهده گردید مدل جمعی تعمیم‌یافته از نظر معیارهای ارزیابی، عملکرد بهتری نسبت به مدل خطی تعمیم-یافته دارد.
In the present study, evaluated predictability of generalized additive and linear models in R software by applying selection variable different method for dominant height of beech species as a high criterion for site productivity. Dominant height defined as average height of three highest trees in any sample plots. For this purpose, locate 127 circular sample plots with an area of 1000 square meters in beech dominated forests in research forest of Tarbiat Modares university and in each of them height and diameter of Fagus Orientalis Lipsky trees that greater than a diameter of 7.5 cm within each of plot was recorded along with elevation and percent slope and azimuth. Also, at the center of each sample plot, soil samples from 0-10 cm depth were taken, and several soil physical and chemical variables were measured. In this study, performance of five variable-selection methods evaluated individually for each of generalized linear and additive modeles. In order to compared the performance of variable-selection methods in generalized linear model, is used cross-validation with 2500 repeated and for generalized additive model is cross-validation 10-fold. After selecting the best method of variable selection in each of generalized linear and additive models, obtained relative importance of any important variable that finally altitude variable explored as the most important effective variable on dominant height of beech species. Then Using the evaluation criteria of data modeling which showed generalized additive model of evaluation criteria,has better performance than generalized linear model.
Refrences
1- Aertsen, W., V. Kint, B. De Vos, J. Deckers, J. Van Orshoven, & B. Muys, 2010. Comparison and ranking of different modelling techniques for prediction of site index in Mediterranean mountain forests. Ecological Modeling, 221(8): 1119-1130.
2- Alder, D, 1980. Forest volume estimation and yield prediction, F.A.O. Publication, Rome.
3- Berges, L., R. Chevalier, Y. Dumas, A. Franc, J.M. Gilbert, 2005. Sessile oak (Quercus petraea Liebl.) site index variations in relation to climate, topography and soil in even-aged high-forest stands in northern France. Annals of Forest Science, 62(5): 391-402.
4- Burnham, K.P., D.R. Anderson, 2002. Model selection and multimodel inference: A practical information-theoretic approach, 2nd edition, Springer, New York.
5- Burnham, K.P., D.R. Anderson, K.P. Huyvaert, 2011. AIC model selection and multimodel inference in behavioral ecology: some background, observations, and comparisons. Behavioral Ecology and Sociobiology 65:23-35.
6- Chatterjee, S., S.A. Hadi, & B. Price, 2004. Regression Analysis by Example. Niroumand, H.A., Publishers University of Mashhad, 384p.
7- Fakoor, E., 2014. Evaluating the site productivity of Oriental Beech using data mining techniques (decision tree). Master's thesis, University of Tarbiat Modares, 79p.
8- Gerritsen, J., J.M. Dietz, & H.T. Wilson, 1996. Episodic acidification of coastal plain streams: an estimation of risk to fish. Ecological Applications, 6(2): 438–448.
9- Hastie, T., R. Tibshirani, 1986. Generalized Additive Model. Statistical sincence, 1(3):297-318.
10- Hastie, T., R. Tibshirani, 1990. Non-parametric logistic and proportional odds regression. Applied statistics, 260-276.
11- Heegaard, E., 2002. The outer border and central border for species–environmental relationships estimated by non-parametric generalised additive models. Ecological Modelling, 157(2): 131-139.
12- Hobbs, N.T., & R. Hilborn, 2006. Alternatives to statistical hypothesis testing in ecology: A guide to self-teaching. Ecological Applications, 16(1): 5–19.
13- Hoeting, J.A., R.A. Davis, A.A. Merton, & S.E. Thompson, 2006. Model selection for geostatistical models. Ecological Applications, 16: 87–98.
14- Jafarihaghighi, M., 2003. Methods of soil analysis, Publication voice of Zoha, 236p, Sari.
15- Kalantari, KH., 2010. Processing and analyzing data in social - economic research, Saba Cultural Publications, 388 p.
16- Knick, S.T., & J.T. Rotenberry, 1995. Landscape characteristics of fragmented shrubsteppe habitats and breeding passerine birds. Conservation Biology, 9(5): 1059–1071.
17- Lee, H., & S. Ghosh, 2009. Performance of information criteria for spatial models. Journal of Statistical Computation and Simulation, 79(1): 93–106.
18- Lehmann, A., J.M. Overton, J.R. Leathwick, 2002. GRASP: generalized regression analysis and spatial prediction. Ecological modelling, 157(2): 189-207.
19- Leigh, G.T., A.J. Read, & P. Halpin, 2008. Fine-scale habitat modeling of a top marine predator: do prey data improve predictive capacity. Ecological Applications, 18(7): 1702–1717.
20- Lukacs, P.M., W.L. Thompson, W.L. Kendall, W.R. Gould, Jr P.F. Doherty, K.P. Burnham, & D.R. Anderson, 2007. Concerns regarding a call for pluralism of information theory and hypothesis testing. Journal of Applied Ecology, 44(2): 456–460.
21- Mazerolle, M.J., 2006. Improving data analysis in herpetology: using Akaike’s information criterion (AIC) to assess the strength of biological hypotheses. Amphibia-Reptilia, 27(2): 169–180.
22- McLintock, T.F., & C.A. Bickford, 1957. A proposed site index for red spruce in the Northeast, Publication of Northeastern Forest Experiment Station, U.S. Upper Darby.
23- Miller, A., 2002. Subset Selection in Regression, 2nd edition, Chapman & Hall/CRC, Boca Raton, Florida.
24- Moisen, G.G., E.A. Freeman, J.A. Blackard, T.S. Frescino, N.E. Zimmermann, & Jr T.C. Edwards, 2006. Predicting tree species presence and basal area in Utah: a comparison of stochastic gradient boosting, generalized additive models, and tree-based methods. Ecological modelling, 199(2): 176-187.
25- Mundry, R., & C.L. Nunn, 2009. Stepwise model fitting and statistical inference: turning noise into signal pollution. The American Naturalist, 173(1): 119–123.
26- Murtaugh, P.A., 1998. Methods of variable selection in regression modeling. Communications in Statistics: Simulation and Computation, 27(3): 711–734.
27- Murtaugh, P.A., 2009. Performance of several variable‐selection methods applied to real ecological data. Ecology Letters, 12(10): 1061-1068.
28- Olden, J.D., & D.A. Jackson, 2000. Torturing data for the sake of generality: How valid are our regression models? Ecoscience, 7(4): 501–510.
29- Raffalovich, L.E., G.D. Deane, D. Armstrong, & H.S. Tsao, 2008. Model selection procedures in social research: Monte-Carlo simulation results. Journal of Applied Statistics, 35(10): 1093–1114.
30- Ramsey, F., & D. Schafer, 2002. The Statistical Sleuth: A Course in Methods of Data Analysis, 2nd edition, Duxbury Press, Belmont, California.
31- Rezaei, A., & A. Soltani, 2003. Introduction to Applied Regression Analysis, Isfahan University Press, 294 p.
32- Sauerbrei, W., P. Royston, & H. Binder, 2007. Selection of important variables and determination of functional form for continuous predictors in multivariable model building. Statistics in Medicine, 26(30): 5512–5528.
33- Sharifi, A.M., 2014. Evaluate the response species Festuca ovina L to some environmental variables using HOF function in catchment's Galandrud, Master's thesis Department of Natural Resources Tarbiat Modares University.
34-Sheather, S., 2009. A Modern Approach to Regression with R, Springer Texts in Statistics, Springer, 393 pp.
35- Schwarz, G., 1978. Estimating the dimension of a model. Annals of Statistics 6:461–464.
36- Skovsgaard, J.P., & J.K. Vanclay, 2008. Forest site productivity: a review of the evolution of dendrometric concepts for even-aged stands. Journal of Forestry,81(1): 13–31.
37- Vanclay, J.K., 1992. Assessing site productivity in tropical moist forests: a review. Forest Ecology and Management, 54(1): 257-287.
38- Ward, E.J., 2008. A review and comparison of four commonly used Bayesian and maximum likelihood model selection tools. Ecological Modelling, 211(1): 1–10.
39- Whittingham, M.J., P.A. Stephens, R.B. Bradbury, & R.P. Freckleton, 2006. Why do we still use stepwise modelling in ecology and behaviour? Journal of Animal Ecology, 75(5): 1182–1189.
40- Yee, T.W., N.D. Mitchell, 1991. Generalized additive models in plant ecology, Journal of Vegetation Science, 2(5): 587-602.