Long Range Dependency of Apparent Temperature in Birnin Kebbi

Similar documents
Stationary Time Series

Licenciatura de ADE y Licenciatura conjunta Derecho y ADE. Hoja de ejercicios 2 PARTE A

Financial Econometrics Jeffrey R. Russell Midterm Winter 2009 SOLUTIONS

Diebold, Chapter 7. Francis X. Diebold, Elements of Forecasting, 4th Edition (Mason, Ohio: Cengage Learning, 2006). Chapter 7. Characterizing Cycles

Box-Jenkins Modelling of Nigerian Stock Prices Data

Vectorautoregressive Model and Cointegration Analysis. Time Series Analysis Dr. Sevtap Kestel 1

14 Autoregressive Moving Average Models

Methodology. -ratios are biased and that the appropriate critical values have to be increased by an amount. that depends on the sample size.

R t. C t P t. + u t. C t = αp t + βr t + v t. + β + w t

STRUCTURAL CHANGE IN TIME SERIES OF THE EXCHANGE RATES BETWEEN YEN-DOLLAR AND YEN-EURO IN

Arima Fit to Nigerian Unemployment Data

Exponential Weighted Moving Average (EWMA) Chart Under The Assumption of Moderateness And Its 3 Control Limits

OBJECTIVES OF TIME SERIES ANALYSIS

Nature Neuroscience: doi: /nn Supplementary Figure 1. Spike-count autocorrelations in time.

- The whole joint distribution is independent of the date at which it is measured and depends only on the lag.

Lecture Notes 2. The Hilbert Space Approach to Time Series

Chapter 16. Regression with Time Series Data

Section 4 NABE ASTEF 232

Chapter 15. Time Series: Descriptive Analyses, Models, and Forecasting

Quarterly ice cream sales are high each summer, and the series tends to repeat itself each year, so that the seasonal period is 4.

DYNAMIC ECONOMETRIC MODELS Vol. 4 Nicholas Copernicus University Toruń Jacek Kwiatkowski Nicholas Copernicus University in Toruń

Modeling Rainfall in Dhaka Division of Bangladesh Using Time Series Analysis.

Testing for a Single Factor Model in the Multivariate State Space Framework

Robust critical values for unit root tests for series with conditional heteroscedasticity errors: An application of the simple NoVaS transformation

Vehicle Arrival Models : Headway

Smoothing. Backward smoother: At any give T, replace the observation yt by a combination of observations at & before T

Nonstationarity-Integrated Models. Time Series Analysis Dr. Sevtap Kestel 1

Solutions to Odd Number Exercises in Chapter 6

Chapter 5. Heterocedastic Models. Introduction to time series (2008) 1

School and Workshop on Market Microstructure: Design, Efficiency and Statistical Regularities March 2011

COMPARISON OF THE DIFFERENCING PARAMETER ESTIMATION FROM ARFIMA MODEL BY SPECTRAL REGRESSION METHODS. By Gumgum Darmawan, Nur Iriawan, Suhartono

Econ Autocorrelation. Sanjaya DeSilva

Time series Decomposition method

Kriging Models Predicting Atrazine Concentrations in Surface Water Draining Agricultural Watersheds

Forecasting optimally

Lecture 5. Time series: ECM. Bernardina Algieri Department Economics, Statistics and Finance

DYNAMIC ECONOMETRIC MODELS vol NICHOLAS COPERNICUS UNIVERSITY - TORUŃ Józef Stawicki and Joanna Górka Nicholas Copernicus University

Unit Root Time Series. Univariate random walk

ACE 562 Fall Lecture 8: The Simple Linear Regression Model: R 2, Reporting the Results and Prediction. by Professor Scott H.

Wavelet Variance, Covariance and Correlation Analysis of BSE and NSE Indexes Financial Time Series

Department of Economics East Carolina University Greenville, NC Phone: Fax:

Choice of Spectral Density Estimator in Ng-Perron Test: A Comparative Analysis

Introduction D P. r = constant discount rate, g = Gordon Model (1962): constant dividend growth rate.

How to Deal with Structural Breaks in Practical Cointegration Analysis

Mean Reversion of Balance of Payments GEvidence from Sequential Trend Break Unit Root Tests. Abstract

Navneet Saini, Mayank Goyal, Vishal Bansal (2013); Term Project AML310; Indian Institute of Technology Delhi

ST4064. Time Series Analysis. Lecture notes

Sub Module 2.6. Measurement of transient temperature

Wisconsin Unemployment Rate Forecast Revisited

Cointegration in Theory and Practice. A Tribute to Clive Granger. ASSA Meetings January 5, 2010

Bias in Conditional and Unconditional Fixed Effects Logit Estimation: a Correction * Tom Coupé

Imo Udo Moffat Department of Mathematics/Statistics, University of Uyo, Nigeria

Institute for Mathematical Methods in Economics. University of Technology Vienna. Singapore, May Manfred Deistler

DEPARTMENT OF STATISTICS

State-Space Models. Initialization, Estimation and Smoothing of the Kalman Filter

20. Applications of the Genetic-Drift Model

ACE 562 Fall Lecture 4: Simple Linear Regression Model: Specification and Estimation. by Professor Scott H. Irwin

Cointegration and Implications for Forecasting

Explaining Total Factor Productivity. Ulrich Kohli University of Geneva December 2015

Introduction to Probability and Statistics Slides 4 Chapter 4

5.2. The Natural Logarithm. Solution

Development of a new metrological model for measuring of the water surface evaporation Tovmach L. Tovmach Yr. Abstract Introduction

Estimation of Poses with Particle Filters

Computer Simulates the Effect of Internal Restriction on Residuals in Linear Regression Model with First-order Autoregressive Procedures

ACE 564 Spring Lecture 7. Extensions of The Multiple Regression Model: Dummy Independent Variables. by Professor Scott H.

Distribution of Least Squares

LONG MEMORY AT THE LONG-RUN AND THE SEASONAL MONTHLY FREQUENCIES IN THE US MONEY STOCK. Guglielmo Maria Caporale. Brunel University, London

STAD57 Time Series Analysis. Lecture 17

STAD57 Time Series Analysis. Lecture 17

STATE-SPACE MODELLING. A mass balance across the tank gives:

BOOTSTRAP PREDICTION INTERVALS FOR TIME SERIES MODELS WITH HETROSCEDASTIC ERRORS. Department of Statistics, Islamia College, Peshawar, KP, Pakistan 2

Regression with Time Series Data

Yong Jiang, Zhongbao Zhou School of Business Administration, Hunan University, Changsha , China

DEPARTMENT OF ECONOMICS AND FINANCE COLLEGE OF BUSINESS AND ECONOMICS UNIVERSITY OF CANTERBURY CHRISTCHURCH, NEW ZEALAND

Forecasting models for economic and environmental applications

A note on spurious regressions between stationary series

GINI MEAN DIFFERENCE AND EWMA CHARTS. Muhammad Riaz, Department of Statistics, Quaid-e-Azam University Islamabad,

Robust estimation based on the first- and third-moment restrictions of the power transformation model

A Specification Test for Linear Dynamic Stochastic General Equilibrium Models

Linear Time-invariant systems, Convolution, and Cross-correlation

Derived Short-Run and Long-Run Softwood Lumber Demand and Supply

Mathematical Theory and Modeling ISSN (Paper) ISSN (Online) Vol 3, No.3, 2013

Some Basic Information about M-S-D Systems

A unit root test based on smooth transitions and nonlinear adjustment

ESTIMATION OF DYNAMIC PANEL DATA MODELS WHEN REGRESSION COEFFICIENTS AND INDIVIDUAL EFFECTS ARE TIME-VARYING

The General Linear Test in the Ridge Regression

Outline. lse-logo. Outline. Outline. 1 Wald Test. 2 The Likelihood Ratio Test. 3 Lagrange Multiplier Tests

Chapter 3, Part IV: The Box-Jenkins Approach to Model Building

Modeling of Sokoto Daily Average Temperature: A Fractional Integration Approach

The Simple Linear Regression Model: Reporting the Results and Choosing the Functional Form

Time Series Models for Growth of Urban Population in SAARC Countries

GDP PER CAPITA IN EUROPE: TIME TRENDS AND PERSISTENCE

Comparing Means: t-tests for One Sample & Two Related Samples

23.2. Representing Periodic Functions by Fourier Series. Introduction. Prerequisites. Learning Outcomes

Y 0.4Y 0.45Y Y to a proper ARMA specification.

ACE 562 Fall Lecture 5: The Simple Linear Regression Model: Sampling Properties of the Least Squares Estimators. by Professor Scott H.

(10) (a) Derive and plot the spectrum of y. Discuss how the seasonality in the process is evident in spectrum.

EE 435. Lecture 35. Absolute and Relative Accuracy DAC Design. The String DAC

Properties of Autocorrelated Processes Economics 30331

Generalized Least Squares

Transcription:

Inernaional Journal of Saisics and Applicaions 017, 7(6): 98-303 DOI: 10.593/j.saisics.0170706.04 Long Range Dependency of Apparen Temperaure in Birnin Kebbi Babayemi A. W. 1,*, Onwuka G. I. 1, Olorunpomi O. T. 1 Deparmen of Mahemaics, Kebbi Sae Universiy of Science and Technology, Aliero, Nigeria Deparmen of Mahemaics, Nigeria Police Academy, Wudil, Nigeria Absrac This research examined he long range dependency compormen of apparen emperaure in BirninKebbi meropolis, using hourly daa for a period of wo years from 013 o 014. We found ha he daa poins are highly correlaed, here is a sense of reurn of a paricular characerisic afer some hours and his depics seasonaliy rais and fracional inegraed d series suspeced. Auoregressive Fracional Inegraed Moving Average (ARFIMA) modelrevealed ha he series for he periods invesigaed is fracionally inegraed wih high-lag correlaion srucure, d=0.495. The long range dependency which describes he number of days for he reurn characerisics of persisence of he shock in a very long ime period afer i occurs was esimaed o be hree hundred and welve days (31) days approximaely a year; hus, once a shock is fel on apparen emperaure, i will go on for abou a year before he paern or srucure will be repeaed. Keywords Apparen emperaure, ARFIMA, High-lag correlaion srucure, Long range dependency 1. Background o he Sudy Apparen emperaure allows for he quaniaive assessmen of hea sress and is used o deermine he limi of hea exposure in order o preven hea relaed illnesses. The Hea Index was originally developed by Seadman (011) as an assessmen for sulriness, bu commonly referred o as he apparen emperaure. Parameers involved in deermining apparen emperaure are waer vapor pressure, surface area of skin, significan diameer of a human, clohing, core emperaure, aciviy, effecive wind speed, clohing resisance o hea ransfer, radiaion o and from he skin s surface, sweaing rae, venilaion rae, skin resisance o hea ransfer, and surface resisance o moisure ransfer. The hea index is he measure of how ho i really feels when relaive humidiy is facored wih he acual air emperaure (Rohfusz, 1990; Mendelsohn, 003; Epsein, 006; Rober, e al. 010; Anderson e al. 013; NOAA, 013; Qu, e al. 016). In esimaion of ARIMA (p,d,q) model he researchers ofen peg he value of d as an ineger value mos especially beween 0 and, however in esimaing d in pracice i is no always an ineger value here are ofen a imes ha he value of d lies beween -0.5 and 0.5. Many ime series exhibi oo much long range dependence o be classified as eiher I(0) or I(1) bu do no belong o eiher of * Corresponding auhor: whafsa@gmail.com (Babayemi A. W.) Published online a hp://journal.sapub.org/saisics Copyrigh 017 Scienific & Academic Publishing. All Righs Reserved hem. In he recen, majoriy of hese long range dependence models are ofen called long memory models are normally aken care of by he Auoregressive Fracional Inegraed Moving Average (ARFIMA) model. The series ha exhibis long memory process has auocorrelaion funcion ha damps hyperbolically more slowly han he geomeric damping exhibied by shor memory auoregressive moving-average, ARMA(p,q) process, which is predicable a long horizons(granger, e al. 1980; Hosking, 1981; Ding, e al. 1993; Taqqu, e al. 1995; Baillie, 006; Robinson, 003; Agosinelli, e al. 003; Larranga, 00; Nesrin, 006; Haldrup, e al 007; Baum, 013; Conreras-Reyes, e al. 013). Prior o his invesigaion on long memory of hea index in BIRNIN KEBBI meropolis, here have been previous sudies involving hea in oher fields such as Agric, Engineering, Physics, Chemisry or Biology in KEBBI Sae bu lile or no enquiry has been carried ou on long memory effec of apparen emperaure in KEBBI using ime series approach. Over he years, here ends o be a severe hea during a paricular period each year in he invesigaed area. This means we have higher values of hea index a a paricular poin in ime every year. The research aims a invesigaing he long range dependency characerisics of apparen emperaure in BIRNIN KEBBI meropolis wih a view o deermining: (i) he auocorrelaion paern for he long range dependency of he daa poins for he research; (ii) fracional value of d for he daa poins; and (iii) inerpre he long memory characerisics.

Inernaional Journal of Saisics and Applicaions 017, 7(6): 98-303 99. Mehodology We adoped ime series mehodology o long memory characerisic of apparen emperaure in BIRNIN KEBBI meropolis. I deals wih hourly daa of hea index colleced over a period of wo years from 013 o 014, and he se of daa would be subjeced o saisical esing procedures o confirm if he daa are ruly long memory, inegraed, or saionary processes. I is expeced o add o he exising lieraure on long memory behavior in hea index and will end o help he merology or praciioners in field of hydrology or dendrology. The daa are secondary and are colleced from Energy Research cener BIRNIN KEBBI..1. ARMA Process An auoregressive moving average (ARMA) process consiss of boh auoregressive and moving average erms. If he process has erms from boh an AR(p) and MA(q) process, hen he process is called ARMA(p, q) and can be expressed as: ( B)( y ) ( B) φ µ = θ ε (1) where εε ~WWWW(0, σσ εε ) is a whie noise disribuion erm, so ha for all, E ( ε ) = 0 ; Var ( ε ) = σε and ε are serially uncorrelaed. All ARMA models are saionary. If an observed ime series is non-saionary (e.g. upward rend), i mus be convered o saionary ime series (e.g. by differencing) (Anderson, e al 1979; Pelgrin, 011)... ARFIMA Process ARFIMA process allows for long memory series o be fracionally inegraed. The general form of fracional ARMA model or he ARFIMA (p,d,q) model is: d ( B)( 1 B) y ( B) φ = θ ε () Where d denoes non-ineger fracional differencing parameer, y is he dependen variable a ime, ε is disribued normally wih mean 0 and variance σ, ε ( ) 1 φ B = φ B φ B φ B 1 p and p ( ) 1 θ B = 1+ θ B p + θb + + θpb represens AR and MA componens respecively; ( B) Φ and ( B) Θ have no common roos, B is he backward shif operaor and ( 1 ) B d is he fracional differencing operaor given by he binomial expansion: ( 1 ) large j is ηη jj Γ ( j+ d) ( j ) ( d) d j j B = B = η B Γ + 1 Γ j j= 0 j= 0 jj dd 1 for ; d ( 0.5 / 0.5) and ( ) Γ(d) ε is a whie noise sequence wih zero mean and Innovaion variance σ. The properies of an ARFIMA process area bridged in he following heorem. Le hen: (a) (b) y be an ARFIMA (p,d,q) process y is saionary and inverible if d < 0.5 and all he roo of φ ( B) = 0 lie ouside he uni circle. y is non-saionary if d 0.5 and all he roo of θ ( B) = 0 lie ouside he uni circle. (c) if 0.5 < d < 0.5, he auocovariance of d y; γ = E yy Bk, as k ; where B k k is a funcion of d. ( ) 1 The auocovariance funcion of ARFIMA process decay hyperbolically o zero as k in conras o he faser geomeric decay of a saionary ARMA process..3. Pormaneau Tes I is a es used for invesigaing he presence of auocorrelaion in ime series, number of lags µ and h are predeermined. Hypohesis HH OO : ρρ μμ,1 =. = ρρ μμ,h = 0 (all he lags correlaing are zero) H i : ρρ μμ,1 0 i.e for a leas one i=1,..h is esed i.e a leas one lag wih non zero correlaion. Tes Saisic LLLL h = TT h ( 1 jj =1 TT JJ ) ρρ μμ,jj = T 1 jj μμ iiii μμ ρρ μμ,jj where 3. Analysis of Daa and Resuls Figure 3.1 reveals he embedded long range dependency characerisics of he hourly daa poins in KEBBI beween 013/14. The daa are highly correlaed, here is a sense of reurn of a paricular characerisic afer some hours and his depics seasonaliy rais and fracional inegraed d series suspeced. There was a drop afer he firs seven housand hours of which i now picks and falls afer fifeen hours, also he graph poin shows he endency of high correlaion beween he se of daa poin which amoun o he fac he series is sable over ime. The auocorrelaion funcion of he firs wo housand five hundred (500) observaions in he daa se displayed a rai of seasonaliy as i can be seen in he figure above ha for every lag muliple of welve (600) here appeared rough and picks, very srong correlaion was noiced. The figure 3.3 revealed ha when he number of he lag is double we have a longer urning poin, very srong correlaion was noiced. The figure 3.4 shows ha here is a very srong correlaion among he daa poins and he sinusoidal movemen imiaes a seasonal rai.

300 Babayemi A. W. e al.: Long Range Dependency of Apparen Temperaure in Birnin Kebbi Figure 3.1. Visual Represenaion of Hourly Hea Index in KEBBI beween 013 and 014 Figure 3.. Visual Represenaion of Firs Two Thousand Five Hundred Hours Observaion Figure 3.3. Visual Represenaion of Five Thousand Hours Observaions

Inernaional Journal of Saisics and Applicaions 017, 7(6): 98-303 301 Figure 3.4. Visual Represenaion of Ten Thousand Hours Observaions Table 3.1. Descripive Saisics for he Hea Index a Differen Hours NOB Mean Sd Median Min. Max. Skewness Kurosis Se 15418 7.7 11.31 8 1 59 0.1 0.87 0.09 500 37.61 8.789 37 13.0 61 0.17.46 0.178 5000 4.59 11.99 43.00 5.0 55.6 15.6 67. 15.6 10000 41.1 11. 41.0 5.0 55.6 9.64 440.0 9.6 Table 3.1 shows he descripive saisics of he hea index where he all samples are sub divided in hree caegories he firs wo housand five hundred observaions, five housand observaions and en housand observaions; in order o examine he characerisics of he inegraed order of a differen sample hours. The minimum value of hea index for he hall sample is 1 and he maximum is 59. The average mean of he hall sample is 7.7. Table 3.. Pormaneau Tes for Auocorrelaion S/N ACF PACF Q-STAT p-value 1 0.093 0.093 13.5 0.000 0.087 0.079 49.36 0.000 3 0.081 0.068 351.68 0.000 4 0.075 0.057 438.68 0.000 5 0.070 0.049 514.3 0.000 6 0.06 0.038 57.69 0.000 7 0.055 0.031 619.3 0.000 8 0.048 0.04 655.35 0.000 9 0.044 0.00 684.64 0.000 10 0.037 0.014 705.6 0.000 11 0.033 0.011 71.8 0.000 1 0.030 0.010 736.1 0.000 13 0.033 0.014 75.9 0.000 14 0.035 0.017 771.98 0.000 15 0.041 0.03 797.84 0.000 16 0.045 0.06 88.96 0.000 17 0.050 0.030 867.43 0.000 18 0.054 0.03 91.44 0.000 19 0.059 0.035 967.04 0.000 0 0.065 0.038 103.0 0.000 1 0.071 0.041 1108.9 0.000 The able 3. above shows he resuls of he Pormaneau es a lag inerval of 00 of which he firs 1 of hem are displayed. I reveals highly correlaed se of daa poins which connoes non saionary level of he series. Table 3.3. ADF Tes for he Long Memory Process Dickey-fuller Saisics Lag-order p-value 6.058 4 0.01 The resul from dickey fuller appears as if he series of apparen emperaure is saionary unforunaely however, i is a long memory process where he value of he inegraed order d in his case is no uni raher, a fracional value. Having searched he ARFIMA models in an inerval of [0,3] for boh orders of p and q wih fracional inegraed d ( 0.5,0.5) using Schwarz Informaion Crierion, he ARFIMA(3, 0.4950, 1) end o be he mos preferable model, he resuls are displayed in Table 3.4 below.

30 Babayemi A. W. e al.: Long Range Dependency of Apparen Temperaure in Birnin Kebbi Table 3.4. Selecion of ARFIMA Models Model Loglikelihood d Sd error p- value ARFIMA(3,1) 1394.57 0.4950 0.00000017 0.00 ARFIMA(3,) 1391.45 0.4991 0.00000049 0.00 ARFIMA(,) 13869.91 0.4998 0.000196000 0.00 ARFIMA(3,3) 1673.58 0.4999 0.011980000 0.00 ARFIMA(,3) 1460.11 0.9960 0.000014500 0.00 Esimaion of he ARFIMA(3, 0.4950, 1) Variable coefficien Sd Error p-value d 0.4950 0.00000017 0.0000 AR(3) 0.1858 0.008396000 0.0000 MA(1) 0.536 0.00867000 0.0000 R-squared=0.86433; Loglikelihood=1394.57; Invered MA roo=-0.5 Invered AR Roos= 0.57,-0.9+50i, 0.9-50i 4. Discussion of Resuls The research invesigaed he long range dependency characerisics of apparen emperaure in Birnin Kebbi using ARFIMA model. The plos and he original auocorrelaion srucure are carefully examined and i is eviden from he ADF es ha he series under invesigaion does no conain uni roo, neiher can i be associaed wih a saionary o a series. I is noiced from he preliminary invesigaion he series possess long range dependency characerisics of d lying beween 0 and 0.5 using Geweke, e al (008) esimaor and ARFIMA. ARFIMA model gives a beer esimae. Long range dependency end o reflec he persisence of shock in a very long ime period afer i occurs, meaning ha; once a shocks is fel on he apparen emperaure, i will go ono a very long ime period for which he paern or srucure will be repeaed. This can also be seen in he auocorrelaion srucure of all samples and when i is sub divided in o hree caegories, he average ime of reurn is hree hundred and welve days which is approximaely one year o all samples. Sample 500 showed asinusoidal paern in he srucure, while samples 500, 5000 and 10000 imiaed he repeiive naure of shock persisence afer a very long period of ime. This shows ha here is always a period where he apparen emperaure is a pick or a lowes poin. This research work preven he misleading inegraed of d = 1 for ARIMA process. 5. Conclusions The research was se ou o invesigae he long memory behaviour of apparen emperaure in KEBBI meropolis, KEBBI Sae using hourly daa for a period of wo years (013/14). Time series mehodology of Auoregressive Fracional Inegraed Moving Average model (ARFIMA) is used in he sudy. The resuls revealed ha he series for he period invesigaed is highly correlaed and fracionally inegraed wih he value of d is 0.495 and he number of days for he reurn characerisics of he persisence of he shock is esimaed o be 31 days approximaely a year. REFERENCES [1] Agosinelli, C. and L. Bisaglia (003). Robus esimaion of ARFIMA process. Technical Repor UniversiàCàFoscari di Venezia. [] Anderson, O. D. and De Gooijer, J. G. (1979). \On discriminaing beween IMA(1,1) and ARMA(1,1) processes: Some exensions o a paper by Wichern," The Saisician, 8,(), Pp. 119-133. [3] Baillie, R.T. (1996). Long memoery process and fracional inegraion in economerics. Journal of Economerics 73; Pp. 5-59. [4] Baum, C.F. (013). ARFIMA (long memory) models, Boson college, www.bc,edu/ec-c/s 013/83/Ec83,s013.nn08.sli de pgf. [5] Conreras-Reyes, J.E. and W. Palma (013). Saisical Analysis of Auoregressive Fracionally Inegraed Moving Average models in R. Springer-Verlag Berlin Heidelberg. arxiv: 108.178v1 [sa.co] 8 Aug 01; doi.10.1007/s00180-013-0408-7. [6] Dickey, D.A. and W.A Fuller (1979). Disribuion of he Esimaors for Auoregressive Time Series wih a Uni Roo. Journal of he American Saisical Associaion, 74, Pp. 47 431. [7] Ding, Z., Granger, C.W.J. and R.F. Engle (1993). A Long Memory Propery of Sock Reurns and a New Model, Journal of Empirical Finance, 1, Pp. 83-106. [8] Epsein,Y., and D. Moran (006). Thermal comfor and he hea sress indices. Indusrial Healh, Pp. 388 398. [9] Pelgrin, F. (011). Lecure : ARMA(p,q) model (par 3). Universiy of Lausanne, _Ecole des HEC Deparmen of mahemaics (IMEA-Nice).

Inernaional Journal of Saisics and Applicaions 017, 7(6): 98-303 303 [10] Geweke, J. and S. Porer-Hudak (1983). The esimaion and applicaion of long memory ime series models. Journal of Time Series Analysis. 4: Pp.1-37. [11] Granger, C. W. J. and R. Joyeux (1980). "An inroducion o long-memory ime series models and fracional differencing". Journal of Time Series Analysis. 1: Pp.15 30. doi:10.1111/j.1467-989.1980.b0097.x. [1] Haldrup, N. andm. Nielsen (007). Esimaion of fracional inegraion in he presence of daa noise. Compuaional saisical and daa analysis, 51: Pp.3100-3114. [13] Hosking, J. R. M. (1981)."Fracional differencing". Biomerika. 68 (1): Pp.165 176. doi:10.1093/biome/68.1.165. [14] Larranga, M. (00). Thermal Sandards and Measuremen Techniques. In D. Anna, The Occupaional Environmen: Is Evaluaion, Conrol, and Managmen. Fairfax: American Indusrial Hygiene Associaion Pp. 918-951. [15] Mendelsohn, R. 003. Appendix XI: The impac of climae change on energy expendiures in California. Global Climae Change and California: Poenial Implicaions for Ecosysems, Healh, and he Economy, C. Thomas and R. Howard, Eds., Sacrameno, California (003). [16] Naional Oceanic Amospheric Associaion (013). Naional Weaher Service Hea Index. Hea Index. NOAA hp://www.nws.noaa.gov/os/hea/images/heaindex.png. [17] Nesrin, A. (006). Long Memory Analysis of USD/TRL Exchange Rae, World Academy of Science, Engineering and Technology Inernaional, Journal of Social, Human Science and Engineering, 3(7): Pp.1-3. [18] Rober, H.S and D.S. Soffer (010). Time series and is Applicaion wih R. Third Ediion. Springer New York Dordrech Heidelberg London. doi 10.1007/978-1-4419-7865-3. [19] Robinson, P. M. (003). Time Series Wih Long Memory. Oxford Universiy Press. ISBN 0-19-9579-9. [0] Rohfusz, L.P (1990). The Hea Index Equaion or, More Than You Ever Waned o Know Abou Hea Index. Technical Aachmen, 1990. Available online: hp://www.s rh.noaa.gov/images/ffc/pdf/ahindx.pdf (Accessed on 15 November 015). [1] Seadman, R.G (011), The assessmen of sulriness. Par 1: A emperaure-humidiy index based on human physiology and clohing science. J. Appl. Meeor. [] Taqqu, M. S.; Teverovsky, V. and W. Willinger (1995). "Esimaors for Long-Range Dependence: An Empirical Sudy". Fracals. 3 (4): Pp.785 798. doi:10.114/s018348x9500069. [3] Qu, L., Chen, J., Dang, G, Jiang, S., Li, L., Guo, J. e al (016). Hea Wave reduces Ecosysem Carbonsink Srengh in a Eurasian Meadow Seppe. Environ Res. 144: 39-48. dio:10.1016/j.envres.015.09.004.