Spatial Disparity Analysis of Inner Mongolia Based on GIS

Transcription

1 2015 AASRI Iteratoal Coferece o Idustral Electrocs ad Applcatos (IEA 2015) Spatal Dsparty Aalyss of Ier Mogola Based o GIS Xaolag Lu, Zhyag Ja, Wal Lu Toursm ad Culture College of Yua Uversty, Ljag, Cha Abstract Wth the applcato of Geographc Iformato Systems (GIS) ad Exploratory Spatal Data Aalyss (ESDA) techques, twelve crtera dcatg the couty ecoomc stregth of Ier Mogola are selected to evaluate a aalyss ad dscussed ts causes o the spatal dsparty. I order to make the factor aalyss, prcpal compoet aalyss (PCA) s performed by the SPSS software. Coclusos cocerg global spatal autocorrelato aalyss ad local spatal autocorrelato aalyss are studed. Some sgfcat graphs ad charts are demostrated accordg to statstcs of the results. Keywords-spatal dsparty aalyss; er Mogola; GIS; ESDA; PCA; SPSS I. INTRODUCTION Regoal dvergece s a commo pheomeo the regoal ecoomc developmet as well as a core topc of regoal ecoomcs. The research scale of the regoal dvergece s relatvely dstct of whch t s volved wth the dspartes amog the easter, cetral ad wester belts, betwee the trastoal coutes ad at the couty scale. Exploratory Spatal Data Aalyss (ESDA) techques are dscussed from the evaluato of sgle crtera to the evaluato of mult-crtera [1]. Sce the couty ecoomy plays a mportat role govermet decso-makg ad macro-cotrol, the research of the dspartes at the couty scale s sgfcat for the provcal govermets ad cetral govermet to cotrol the regoal ecoomc developmet [2]. May scholars have appled GIS techology to speces dstrbuto, demographc chage, botay, regoal ecoomy [3], etc. II. METHODOLOGY O the bass of the admstratve dvso of Ier Mogola 2010, we chooses 101 coutes (baer, cty ad dstrct) Ier Mogola as our study objects, s supported by GIS software platform [4] ad takes the SPSS statstcal software as the auxlary tool. Wth the spatal autocorrelato approach, the frequecy map, global ad local autocorrelato testg ad other tools, the quattatve research s coducted o the spatal dspartes of the couty ecoomc stregth of Ier Mogola autoomous rego. A. Selecto of Evaluato Crtera O accout of the relatvely comprehesve ature of the mult-crtera evaluato, the multple crtera are appled to evaluate the couty ecoomc stregth ths paper. The couty ecoomc stregth s maly measured by the ecoomc developmet level at per capta dex, reflectg the dustral ecoomc structure ad the ecoomc beeft of output-put rato, etc. The specfc dexes are stated as follows: Per Capta Reveue (x1), Average Wages of O-Post Staff (x2), Rural Per Capta Net Icome (x3), Per Capta Depost Balace (x4), Proporto of Output Value of the 2d Idustry (x5), Proporto of Output Value of the 3rd dustry (x6), Proporto of Employed Persos of the 2d Idustry (x7), Proporto of Employed Persos of the 3rd dustry (x8), Rato of fscal Reveue to GDP (x9), Idustral Value- Added Rate (x10), Per Capta GDP (x11), Total Retal Sales of Socal Cosumer Goods (x12). Amog the twelve crtera, the ut of total retal sales of socal cosumer goods s oe hudred mllo CNY. B. Factor Aalyss The SPSS software s appled to make the factor aalyss of whch the ma approach s Prcpal Compoet Aalyss (PCA) [5] makg the orgal mult-crtera x tegrated to oe or several dexes that the small amout of dexes wll reflect the vast majorty of formato of mult-crtera. C. Spatal Autocorrelato Aalyss Spatal autocorrelato, as oe of the mportat forms of spatal depedece, refers to the correlato betwee the research object ad the spatal locato. Spatal autocorrelato s a mportat dex to verfy whether there s a sgfcat assocato betwee a partcular attrbute value ad the attrbute value of ts adjacet elemet. Wth the assstace of the Arcvew software, t s possble to aalyze the spatal autocorrelato. D. Data Source The selected crtera could be rather complete to reflect the couty ecoomc stregth. Sce the formato s acqured from the curret offcal statstcs data wth the good credblty ad tegrty. The data orgates from the statstcal yearbook of Ier Mogola. The spatal aalyss scale s 101 coutes (baer, cty ad dstrct) the Ier Mogola autoomous rego, ad the admstratve boudary data s take from the admstratve dvso map of Ier Mogola at the proporto of 1: based o vector quatzato of scag the tracks The authors - Publshed by Atlats Press 500

2 III. COUNTY ECONOMIC STRENGTH EVALUATION The prcpal compoet aalyss (PCA) approach s carred out by SPSS. We ca fd that KMO=0.7 amog the 12 crtera, whch dcatg the correlato amog the varables ad the rghtess of makg the factor aalyss. Accordg to the electo prcple that the prcpal dvsor of the egevalues s greater tha 1, the selected egevalues were respectvely 4.307, ad At the same tme, the cumulatve varace cotrbuto s %, whch the majorty of formato s cluded (Table I). Based o the clearer compoet matrx (Table II) obtaed from the orthogoal rotato of the varmax method, the frst prcpal compoet F1 s called after ecoomc developmet factor, the secod prcpal compoet F2 s called after ecoomc structure factor ad the thrd prcpal compoet F3 s called after ecoomc beeft factor. TABLE I. FACTOR EIGENVALUES AND CONTRIBUTION RATE No. Egevalue Cotrbuto rate (%) Accumulatve cotrbuto rate (%) By usg the regresso method [6], we ca calculate each factor score ad ecoomc stregth score F, whch the respectve cotrbuto rates of the frst prcpal compoet F1, the secod prcpal compoets F2 ad the thrd prcpal compoet F3 are ther correspodg weghts of the three prcpal compoets. The cotrbuto rate of the frst prcpal compoet F1 s wth the cotrbuto rate of the secod prcpal compoet F2 at ad the cotrbuto rate of the thrd prcpal compoet F3 at I ths way, the ma-made factor o the scorg prefereces of the experts s restrcted. Just gve the pure umber, the scorg model of the couty ecoomc stregth shall be F= *F *F *F3. We ca calculate the ecoomc stregth score for the coutes of Ier Mogola autoomous rego. Wth the hgher scores, the couty has the hgher level of ecoomc stregth, otherwse vce versa. The fact that the score s greater tha 0 dcates that the level of ecoomc stregth s hgher tha the average level of the whole autoomous rego, ad the score s less tha 0 dcates that the level of ecoomc stregth s lower tha the average level. 501

3 TABLE II. MATRIX OF PRINCIPAL COMPONENT ANALYSIS Varables 1st 2d 3rd Per Capta Reveue (X1) Average Wages of O-Post Staff (X2) Rural Per Capta Net Icome (X3) Per Capta Depost Balace (X4) Proporto of Output Value of the 2d Idustry (X5) Proporto of Output Value of the 3rd dustry (X6) Proporto of Employed Persos of the 2d Idustry (X7) Proporto of Employed Persos of the 3rd dustry (X8) Rato Of Fscal Reveue To GDP (X9) Idustral Value-Added Rate (X10) Per Capta GDP (X11) Total Retal Sales Of Socal Cosumer Goods (X12) TABLE III. FREQUENCY DISTRIBUTION CHARACTERISTICS OF THE ECONOMIC STRENGTH SCORES Egevalues Value Meda Mode Std. Devato Skewess Kurtoss Rage Mmum Maxmum The frequecy dstrbuto characterstcs based o the scores of the couty ecoomc stregth represet the ecoomc stregth for the 101 coutes Ier Mogola. From Fgure 1 ad Table III, we ca fd that the frequecy dstrbuto based o the couty ecoomc stregth score of Ier Mogola has a postve skewess wth the larger value of skewess at of whch the data s devated from the cetral le wth the obvously rght skewed dstrbuto. The result shows that the coutes wth the lower scores are overwhelmg. Table 3 llustrates the kurtoss coeffcet s , whch t shows that the gradet of the frequecy dstrbuto s lower ad eveer tha that of Gaussa dstrbuto. The low-lyg dstrbuto also suggests that the coutes wth the hgher scores occupy a larger proporto. Table 3 llustrates the meda s whch s slghtly less tha the mea value 0. The rage s greater at The value of mode s wth the relatvely greater stadard devato at Fgure 1. Frequecy dstrbuto of the couty ecoomc stregth scores The comprehesve scores of each couty ecoomc stregth ad GIS graphc data are lked together. Through the GIS-based statstcal aalyss, the couty ecoomc developmet levels from 101 coutes Ier Mogola are classfed to four dfferet types (Fgure 2): developed areas (The ecoomc stregth score s more tha 0.7), the secodarly developed areas (The ecoomc stregth score s greater tha 0.5 ad less tha 0.7), the less developed regos (The ecoomc stregth score s greater tha 0.2 ad less tha 0.5) ad the udeveloped area (The ecoomc stregth score s below 0.2). 502

4 Fgure 2. Spatal dstrbuto maps of the couty ecoomc stregth scores IV. SPATIAL AUTOCORRELATION ANALYSIS A. Spatal Weght Matrx Durg the process of coductg the spatal autocorrelato aalyss [7], above of all, t eeds to defe the mutual adjacecy relato of the spatal object ad determe the weght of each space ut so as to uderstad the spatal lkage of the related data the GIS database. The defto of the spatal weght matrx depeds o the adjacecy rule ad the dstace rule. I ths paper, the adjacecy rule s adopted. Followg the adjacecy rule, a bary symmetrc spatal weght matrx W s used to express the spatal adjacecy relatoshp the posto. Whe ad j s adjacet, the equato of the elemet the spatal weght matrx s W=1, otherwse W=0. B. Global Spatal Autocorrelato Aalyss The purpose of global spatal autocorrelato aalyss s to explore the spatal dstrbuto characterstc of the attrbute data values the overall area. Through the global spatal autocorrelato statstcs of the value estmate of Global Mora s I [8], t s avalable to aalyze the spatal correlato ad spatal dsparty degree the overall area. I(d) = S j 2 Wj( x x) j W j Where 2 1 s = ( x x) 2, x deotes the attrbute value at I, ad x s the x arthmetc mea value, W j s spatal weght value. The value rage of Mora's I statstcs s geerally from -1 to 1. If I(d)<0, the t meas that the space has a egatve correlato, dcatg the sgfcat spatal dsparty of the ecoomc stregth (1) betwee the rego ad ts surroudg areas. As the value s approachg to -1, the overall spatal dsparty s greater. If I(d)>0, the the space has a postve correlato, dcatg that the accumulatve degree s sgfcat for the areas wth the hgher (lower) ecoomc stregth. As the value s approachg to 1, the overall spatal dsparty s smaller. If I(d)=0, the t meas the space s ot assocated. If C>1 (Geary coeffcet), the t has a egatve correlato. If C=1, the t meas the there s o assocato. If C<1, the t meas the postve correlato. By (1), the value of Mora's I of the global autocorrelato coeffcet s terms of the couty ecoomc stregth the Ier Mogola autoomous rego, whch the value s greater tha 0 whle the Geary coeffcet s less tha 1 at The result shows that the couty ecoomc stregth of Ier Mogola has a postve correlato. For the Mora dex [9], a stadardzed statstcs Z ca be appled to test the spatal autocorrelato the 101 coutes (ctes, dstrcts) of Ier Mogola. If the correlato of Z s postvely sgfcat, t suggests there s postvely spatal autocorrelato. As well, the observed value (hgh value or low value) teds to be spatal agglomerato. Whe the correlato of Z s egatvely sgfcat, t suggests there s egatvely spatal autocorrelato. The dstrbuto of the smlar observatos teds to be scattered, such as the agglomerato of the hgh ad low observato value. If Z equals 0, the the observed value s depedetly ad radomly dstrbuted. By our calculato, Z= s obtaed. The result (as show Table IV) dcates that there s postvely spatal autocorrelato for the couty ecoomc stregth Ier Mogola, amely the area wth the hgh ecoomc stregth or the area wth the low ecoomc stregth teds to be the spatal agglomerato. TABLE IV. GLOBAL STATISTICS Statstcs Expectato Z-value Mora's I Geary's C G-statstcs e

5 C. Local Spatal Autocorrelato Aalyss results are lked wth the correspodg graphcal data to draw the LSIA map ad geerate the dstrbuto dagram of the local spatal autocorrelato coeffcet (As show Fg. 3) based o the GIS thematc mappg fucto. Fgure 3. Dstrbuto of local spatal autocorrelato coeffcets The local spatal autocorrelato aalyss majorly focuses o the dstrbuto patter of the attrbute value of each ut the heterogeeous space so as to measure the local spatal correlato degree betwee each rego ad ts surroudg areas. The commoly used statstcs s the Local Mora's I, measured by I = Z W (2) j j jz j Where Z ad Z j are the stadardzed forms of the elemet attrbute values, ad W j s spatal weght matrx. If I s sgfcatly greater tha 0, t meas that the space s slghtly dfferet betwee the area ad the surroudg areas. If I s sgfcatly less tha 0, t meas that the space s sgfcatly dfferet betwee the area ad the surroudg areas. (2) s used to calculate Local Mora's I of the local spatal autocorrelato coeffcet for each couty ut of Ier Mogola. Afterwards, the calculated REFERENCES [1] Gulla R, Le Gallo J. Agglomerato ad dsperso of ecoomc actvtes ad aroud Pars: a exploratory spatal data aalyss. Evromet ad Plag B: Plag ad Desg, 37(6) , [2] Hual Xe, Chh-Chu Kug, Yulua Zhao. Spatal dspartes of regoal forest lad chage based o ESDA ad GIS at the couty level Bejg-Taj-Hebe area. Froters of Earth Scece, 6(4), pp , [3] Davd M. Theobald,Do L. Steves Jr.,Des Whte,et al. Usg GIS to Geerate Spatally Balaced Radom Survey Desgs for Natural Resource Applcatos. Evrometal Maagemet, 40(1), pp , [4] J Luo, Da-Hog Wag, et al. Popularzed Spatal Iformato Servce Platform Based o Mult-Major GIS, Secod Iteratoal Workshop o Kowledge Dscovery ad Data Mg, pp , [5] Lhua Zhao, M. Ucles, G. Gregory, ull Hogwe Lu, A. Ma. GIS ad spatal formato busess decso support: the case of automoble servce etwork plag, Iteratoal Coferece o Servces Systems ad Servces Maagemet, pp Vol. 2, [6] Guagqg Ch,Ju Zhu. Spatal Regresso Models for Demographc Aalyss.Populato Research ad Polcy Revew, 27(1), pp 17-42, [7] Che Fe, Du Daosheg. Applcato of tegrato of spatal statstcal aalyss wth GIS to regoal ecoomc aalyss. Geospatal Iformato Scece, 7(4), pp , [8] L Asel. The Mora scatterplot as a ESDA tool to assess local stablty spatal assocato, Spatal Aalytcal Perspectves o GIS, Taylor & Fracs Ic, [9] Yag Yag, Kev K. F. Wog. Spatal Dstrbuto of Tourst Flows to Cha's Ctes. Toursm Geographes: A Iteratoal Joural of Toursm Space, Place ad Evromet, 15(2), pp ,