VNU Joum al of Science, E arth Sciences 25 (2009) 65-75
A pplication o f th e Principal C om ponent A nalysis to explore the relation b etw een land use and solid w aste generation in
the D uy T ie n district, H a N am P rovince, V ietnam
Pham V an C u 1’*, Philippe Charrette2, D inh Thi D ie u 1, Pham N goe H a i1, Le Q uang T oan3,
1International C entre f o r A d va n ced Research on G lobal Change, VNU H anoi 2Unviversité d u Q uẻbec à Montréal.
3Institute o fS p a c e Technology, Vietnam A cadem y o f Science a n d Technology VAST
Received 09 July 2009; received in revised form 22 Júly 2009
A bstract. The paper presents and discusses the methodology used and the results obtained by the application o f the Principal Component Analysis (PCA) on a set o f socio-economical and land use data collected in the D uy Tien district (Ha Nam province), Vietnam. Objective o f this study is to use PCA as a data reduction method to verify if a relation couỉd be established between the quantities o f waste generated in a region and its land use and socio-economical characteristics.
Data was collected by a team from the Center for Applied Research in remote sensing and GIS (CARGIS) at the University o f Sciences in Ha Noi. This study is part o f the research Prọịect
“Study the ỉand use changes and its in/luences to the waste in rural sector o f D uy Tien District based on Remoíe Sensing and GIS u t i ỉ i z a t i o nThe prọịect is funded by Vietnam National University for the period 2007-2009.
Due to the limited availability o f statistic data onỉy three types o f economic activity relation are preỉiminarily chosen for PCA to reveal vvhich activity is the predominant for each commune:
Non-farming income/Agriculture Dimensions, Development o f the Tcniary Sector/Agricuỉture and Non-farming Income /Built-up zones expansion. The quantity o f vvasle is than com pared with the activity identiíìed as predominant. All these results are than im poncđ to G1S environment to give the cartographic presentation and to serve the fiiture anaỉysis.
Keywords: Principal Com ponent Analysis; Waste; Land use change; Economic activity; GIS.
1. In tr o d u c tio n
The p a p e r p re s e n ts a n d d isc u sse s the m eth od olog y u sed a n d th e re su lts o b tain ed by th e ap p licatio n o f th e P rin c ip a l C o m p o n e n t A nalysis (P C A ) o n a s e t o f so cio -eco n o m ical and land u se d a ta c o lle c te d in th e D uy T ien
CoTcsponding author. Tel.: 84-913300970.
E-rnail: pvchanoi@vnn.vn
65
d istrict (H a N a m p ro v in ce ), V ietn am . H a N am p ro vince is a ru ral a re a lo cated abou t 60 k ilo m eters so u th o f H a N o i, th e n atio n al Capital o f V ietnam . T h e D u y T ien d istrict has 19 rural co m m u n es an d tw o tovvns. T h e ru ral com m u nes all to g e th e r c o u n te d 32 61 7 h o u seh o ld s spread in 139 v illag es b ack to y e a r 2 0 0 6 . A t th e sam e tim e , th e av erag e p o p u la tio n o f a D uy Tien d istric t’s co m m u n c w a s 7 195 p erso ns.
66
p.v.
Cu et a i / VN U Ịourtial o f Science, Earth Sciences 25 (2009) 65-75D a ta w as co llected by a team from the C e n te r fo r A p plied R esearch in rem o te sensing a n d G IS (C A R G IS ) at th e U n iv ersity o f S c ie n c e s in H a N o i. T h is stud y is p art o f the re se a rc h P rọịect “S tu d y th e la n d use changes a n d its in /lu e n c e s to th e w a ste in ru ra ỉ s e c to r o f D u y Tien D islric t b a s e d o n R em o te S e n sin g a n d G I S u tìliz a tio n ”. T he p rọ ịec t is funded by V ie tn a m N atio n al U n iv ersity fo r th e period 2 0 0 7 -2 0 0 9 .
T h e m ain o b ịe ctiv e o f o u r stu d y w as to use P C A a s a d ata red uction m eth o d w ith the in ten tio n to v e rily i f a rela tio n co u ld be es ta b lis h e d betvveen th e q u a n titie s o f w aste g e n e ra te d in a region an d its land use and socio- ec o n o m ic a l ch aracteristics. S P S S w as the sta tistic a l so ftw are used to p eríò rm th e PC A .
A íir s t co llectio n o f an aly sis results is p resen te d , describ ed an d d iscu ssed h ere in d e p th fo r ap p licatio n p u rp o se. T h e co m p o n en ts e x tra c te d fo r the case stu d y d escrib e tw o d im e n s io n s o f th e p resen t situ atio n o f D u y T ien d istric t: le v e l o f im p o rta n ce o f n o n -fa rm ìn g in c o m e an d a g ricu ltu re. T w o o th er series o f re s u lts are also b rie íly o u tlin ed an d discussed.
T h e íĩrst c a se ex p o ses o n c e m ore tw o d im e n sio n s: th e d ev elo p m en t o f th e teríia ry s e c to r an d agricu ltu re. F in ally , tw o dim ensions w e re as w ell ex tra cted fo r th e last c a se study:
th e le v e l o f im p ortan ce o f n o n -fa rm in g incom e a n d th e b u ilt-u p zo n e exp a n sio n .
T h e presen tatio n o f th e resu lts is m ain ly b a se d o n carto g rap h y o f th e fa c to r scores p ro d u c e d as a resu lt o f th e ap p licatio n o f the P C A o n th e dataset.
2 . M e th o d o lo g y a n d d a t a
A s m en tio n ed ab o v e, o b je ctiv e o f o u r study is to v e riíy i f th e re ex ists a rela tio n b etw een the q u a n titie s o f w aste g en era ted in a reg io n an d its la n d use an d so cio -eco n o m ical ch aracteristics.
T h is stu d y q u estio n is based o n th e fact th at the
in c re ase o f q u an tity o f vvaste is c o n seq u en c e o f d em o g ra p h ic and e c o n o m ic grow th (C h ristian Z u rb riig g 2 0 0 2 ; A u ro b in d o O g ra 2003; Đ ào T h ắm 2 0 0 7 ). In th e c o n te x t o f D uy T ien vvhere th e e c o n o m ic d ev elo p m en t level o f 19 co m m u n es an d 2 to w n s is qu ite d iíĩe re n t, it is im p o rtan t to ev a iu a te th e im portance o f certain key fa c to rs in th e ir e c o n o m ic activ ities a n d to v e riíy th e rela tio n o f th e se driving facto rs w ith th e w aste qu an tity . T h o se relations are non farm in g in co m e/ag ricu ltu re, tertiary secto r/ag ricu ltu re an d non ía rm in g in c o m e/b u ilt-u p zo n e ex p an sio n . T h e statistic d ata w e use in th is p a p e r are provided b y the D e p a rtm e n t o f N atu ral R esources and E n v iro n m en t an d th e D ep artm en t o f A g ricultu re o f D uy T ie n d istrict.
In th is stu d y P C A is th e m ain too l to seek such a lin e a r c o m b in atio n o f variables in vvhich th e v aria n ce ex tra c te d fro m the v aria b les is m ax im al. It th e n ta k e s aw a y th is v ariance from th e m o d el and trie s íìn d in g a seco n d linear co m b in atio n vvhich co u ld ex p lain th e m axim um p ro p o rtio n o f th e re m a in in g variance, and the p ro cess co n tin u es until all th e variance is ex tra cted (A g ile n t T e c h n o lo g ie s 2005; M . M cA d am s an d A . D em irc i 2 006). T h is is called th e p rin cip al ax is m e th o d and results in o rth o g o n al (th u s un co rrelated ) dim ension rep resen tin g th e se d riv in g factors w hich w e use to a n a ly z e an d in terp rete th e statistic d ata o f D uy T ie n . T h is ap p ro ac h is vvidely used in land use a n a ly sis (Jan P e te r L essch en , P eter H.
V e rb u rg e t al. 2 0 0 5 ).
P e rfo rm in g P C A vvith h elp by SPSS w as a ttem p ted h ere w ith th e aim o f red ucin g the large n u m b e r o f o rig in al v aria b les available (m o re th a n 150) to a s m a lle r n u m b e r o f factors fo r m o d e lin g and in terp retation purposes. In D uy T ie n stu dy th e re is ra th e r sm all num ber o f cases in th e av ailab le d ataset (N=21 c o rre sp o n d in g to 19 co m m u n es and 2 to w n sh ip ). T h e re ío re , o n ly a reduced num ber o f v aria b les co u ld be u sed a t a tim e to allow the PC A to p ro d u ce sig n iíìc a n t results.
p .v . Cu et al. / V N U Ịoum al o f Science, Earth Sciences 25 (2009) 65-75 67
T h e “ sam pling ad eq u a cy ” m easu red thought th e K aiser-M ey er-O lk in (K M O v arie s from 0 to 1) statistic w as also taken into co n sid eratio n in th e v ariab le choice. W e used S P S S to calculate a g lo b al K M O alo n g w ith a n in divid ual K M O for cac h variable inclu d ed in th e P C A . It is g en erally reco gn ized in th e literatu re that o v erall K M O sh o u ld b e 0 .6 0 o r h ig h e r to p ro cee d w ith any fa c to r an aly sis, inclu din g PCA (V in es 2000 ). T h e in d iv id u al K M O have been used to d e te rm in e vvhat variab les to exclư de from th e an aly sis b y d ro p p in g the v aria b le w ith the sm a lle st K M O an d re-ru n n in g th e P C A until a satisfac to ry g lo b al K M O is o b ta in ed (M ark etin g D ep t. S P S S Inc. 20 00).
O n c e these m a th em atical c o n strain ts w ere fulfilled and an ac c e p ta b le so lu tio n w as reach ed, decisions h ad to be m a d e reg ard in g the facto rs to retain in th e an aly sis. T h e m ain crite ria used have b een th e K a ise r’s rule i.e.
th e co m p o n en ts reta in ed w e re th e o n e having eig env alu es strictly g re a te r th a n 1. E m phasis w as also put on th e c o m p re h e n sib ility o f the factors. In o th er w o rd s, th e c o m p o n en ts kept w ere to those vvhose đ im en sio n o f m ean in g w as readily com prehensible in the scope o f the research.
F inally th e fa c to r sc o re s in ta b u la r ỉò rm at w h ere ex po rted fro m S P S S to E X C E L and saved as a D B F (d B a s e IV íò rm a t) file. The resu lting d ataset w a s im p o rted into A rcG IS . A
jo in w as c re a te d betw een th e ad m in istrativ e div isio n s (c o m m u n es) g eo g rap h ic la y er and th is tab u lar d atase t in o rd e r to spatially re p re se n t the o u tcom e o f th e PC A an d d etect p o te n tial sp atial d istrib ution p attem s. T he sy m b o lo g y used to
“spatialize” th e facto r sc o re s w a s b ased on graduated c o lo rs w hich sy m b o lizes th e lo w e r (<
0) facto r sc o re s by co ld co lo rs vvhile w arm co lo rs acc o u n ted fo r th e h ig h est sco res. T h e natural b reaks (“Jen k s” ) m eth o d p ro p o sed by A rcM ap w a s uses crea te th e c la sse s. T h is m ethod selec ts th e c la ss b reak s th at b est g ro u p sim ilar v alu es and m axim ize th e d iffere n ces b etw een c la sse s. E ach co m p o n en t w a s sin gly m apped u s in g d istin ct A rcM ap prọịects.
In th e n e x t p arag rap h s w e vvill p re se n t the resu lts o f an a ly sis o f th e th ree c a se stu d ie s and to shorten th e te x t w e w ill sk ip in term ed iary step s o f calcu la tio n o f such in d icato rs as K M O m easure, C h i S q u are T est, p value.
3 . R e s u lts a n d in te r p r e ta tio n
3.1. C ase s tu d y 1: N o n -fa r m in g in c o m e / A g ric u ltu re D im ensions
A ll th e d a ta u sed fo r th is c a se stu d y are co llected fo r th e y e a r 2006. T he v a ria b le s used to perform th e an a ly sis o f th is c a se stu d y are su m m a rized in T a b le 1 below .
Table 1. Description o f variables used for analyzing Non-farming income/Agriculture Dimensions
Variabỉe Label (english) Description
B e p r o m c u i wood_cooker Number o f wood or straw cookers íòund in the commune.
Number o f household involved in the industrial or small industry
@06Ho_CNTTCN In d u stry jih
sector.
@06DT_lua rice area Land area dedicated to rice crop (paddy íìeld) [ha].
D c h u y e n d u n g p u b lic s e rv iv e a re a Land are used for public infrastructure (e.g. roads) [ha].
@06Ho_CNXD IC income Number o f household with major income from industry or construction.
Agriarea* agri_area_pc Percentage o f total area dedicated to agriculture [ha].
*THs íĩeld was calculated based on existing variables @06Dat_SD (total agriculture dedicated area) and Diei tich (total area).
68
p.v.
Cu et al. / V N U Ịoum al oỊSàence, Earth Sàences 25 (2009) 65-75T h e o u tc o m es o f th e PC A co m p u ted by S P S S are in T ab le 2. C o rrela tio n M atrix (a below : Table 2. Correlation Matrix(a)
wood_cooker Indusứy hh ricc area public_servive area IC incomc agriarea_pc
Correlation vvoodcooker 1.000 -.012 .846 .490 .381 .378
Industryhh -.012 1.000 .165 .647 .887 -.496
ricearea .846 .165 1.000 .711 .496 .394
public servive area .490 .647 .711 1.000 .804 -.142
IC incom e .381 .887 .496 .804 1.000 -.297
agri_area_pc .378 -.496 .394 -.142 -.297 1.000
• Determinant = 0.001
Table 3. Totaỉ Variance Expỉained Component Initial Eigenvalues
Total % o f Variance Cumulative %
1 3.227 53.785 53.785
2 2.032 33.868 87.654
3 .404 6.735 94.389
4 .240 3.996 98.385
5 .068 1.128 99.513
6 .029 .487 100.000
Exlraction Method: Principal Component Analysis T h e T a b le 3 p resen ts th e eig en values c a lc u la te d by S P S S shovving th a t th e first and th e seco n d co m p o n en ts h av e eig en values g re a te r th a n 1, i.e., 3 .22 7 an d 2.0 3 2 relatively.
T h e first co m p o n en t p rovid es 53.8% o f the v aria n ce o f th e d atase t vvhile th e second c o m p o n e n t lak es 3 3 .9 % o f th e variance. H ence, th o s e tw o first co m p o n en ts rep resen t alm ost
88% o f th e total v arian ce ex istin g in th e d ata.
A s p e r K aiser’s ru le o nly th e tw o first c o m p o n e n ts w ere ex lra cted b y S P S S fo r the d a ta se t. A va rim a x ro tatio n w as p eríò rm ed in o rd e r to m ake facto r lo ad in g s o f eac h variable to be m o re clearly d iffere n tiated by factor. An o b liq u e o b lim in rotation w a s also perform ed a fte rw ard s in o rd er to g en era te th e factor c o rrelatio n m a ừ ix w h ich d isp la y s th e the PearsorT s r c o e íĩic ie n ts b ctw een both c o m p o n en ts. B ecau se th is m eth o d looks a fte r a no n -o rth o g o n al (o b liq u e) so lu tio n , th e p urpose
o f is th is o p eratio n w a s to v erify if a p oten tially sig n ifican t co rrelatio n ex ists betw een the c o m p o n en ts. T h e ro tated co m p o n en t m atrix alo n g w ith th e co m p o n en t p lo t allow s d istin g u ish in g th e tw o co m p o n en ts ex tra c te d by th e P C A as sh o w n in T a b le 4.
Table 4. Reproduced Component Matrixes Component M atrừfa)
Component
1 2
public servive area .931 -.053
IC income .914 -.316
rice area .779 .580
Industry hh .710 -.637
agri area_pc -.094 .867
wood cooker .637 .659
Extraction Method: Principal Component Analysis.
(a) 2 components extracted.
Rotated Component Matrixfaỉ
Component
1 2
In d u sữ y h h .951 -.071
IC income .917 .307
public servive area .770 .527
nce area .262 .935
wood cooker .103 .911
agn_areaj>c -.604 .630
Exừaction Method: Principal Component Analysis.
Rolation Method: Varimax vvith Kaiser NormalizatiorL (a) Rotation converged in 3 iterations.
p.v.
C u et aỉ. / VN U Ịournaỉ o f Science, Earth Sciences 25 (2009) 65-75 69T h e first co m p o n en t in clu d es the v ariables rclated to non a g ric u ltu re -re late d revenues. T his co m p o n en t carries th c n u m b e r o f ho useholđs w h o se principal incom e co m es from o th er w o rk m an sh ip s such as m an u al lab ors relatcd to co n stru ctio n . T h e n u m b e r o f h o u seh o ld being a c tiv c in sm all in d u stries is also part o f that co m p o n en t. It is also co n c e rn e d by the in ữ astru c tu res sin ce it in clu d es th e variable p u b ỉic servive a r e a . T h e ex iste n c e o f public infrastru ctu res like p avcd ro ad s m ay stim ulate th e dev elo p m en t o f the in d u strial secto r w hich in tu m provides n o n -fa m iin g rev en ues. T hus, th is íìrst d im en sio n c an be seen as a relative m easu re o f the relativ e im p o rtan ce o f non- ía rm in g incom es in a g iv e n co m m u n ity .
T h e second c o m p o n c n t g ath ers th e variables that rcílect a g ric u ltu rc -re late d w ay o f living such the use o f vvood o f (ric e ) straw fueled d ev ices for co o k in g p u rp o se. P eo p le ten d to use vvood o f straw c o o k e r w h e re th e se resou rces are abundant. A hig h p erce n tag e o f area d ed icated to ag ric u ltu re as vvell as the im portance o f th e n et area used fo r rice paddles are also ind icato rs o f hig h lev els o f agricultural activitics. T he se c o n d c o m p o n e n t can hen ce be seen as an in d icato r o f th e relativ e im portance that th e ía rm in g s e c to r o c c u p ie s in th e local econom y.
F a c to r S co res M a p p in g
A lso called c o m p o n e n t s c o r e s in PC A , facto r scores a re e s tim a tio n s o f th e aclual values o f in dividual c a se s (o b serv atio n s) fo r the com ponents. T h e y a re co m p u ted by ta k in g the c a se ’s stan d ard ized sc o re on each variable, m ultiplied by c o rre s p o n d in g fa c to r loading o f the variable for th e given facto r, and sum these products.
T h e in d iv id u al ĩa c to r sc o re s have been com puted fo r co m p o n e n t 1 “N o n -farm in g incom e” an d c o m p o n e n t 2 “ A g ricu ltu re” and
m appcd to sh o w th e sp atial d istrib u tio n o f the scores. T h e m ap p in g also d isp lay s th e estim ated q u an tity o f w a ste (a s k ilo g ram s p e r p erso n per m onth ) g en era ted in eac h co m m u n e. T h is aim s to help v isu ally p erce iv in g th e re la tio n sh ip ( if an y) betw een e a c h ex tra cted co m p o n en t an d the w aste p ro d u ctio n in th e D uy T ien d istrict. The m aps are p resen ted in th e A p p en d ix A.
“N o n -fa rm in g in co m e ”
T h ere are sev en co m m u n es vvhere th e score fo r “N o n -farm in g in co m e” facto r is g re a te r than zero. T h u s, in th e se co m m u n es (H o à n g Đ ông, Y ên Bắc, D u y M inh, M ộc N am , C h u y ê n N goại and C h âu G ian g ) th e non-farm in co m e vvas m ore im p o rtan t than in th e “ av erag e” co n d itio n s o f the D uy T u y d is tric t in 2006. T h is d o e s n ’t m ean th at the h ou seh o ld s in th e se co m m u n es did not gain a n y in com e from ag ric u ltu re . T his resu lt solely m e an s th at co m p ared to th e rest o f the d istrict th e seven co m m u n es c o u n t m ore ho u seh o ld s vvhich w ere p ro v id ed w ith a non agricu ltu re related incom e. It sh o u ld be n oliced that national ro ad s p ass o v er five c o m m u n es for vvhom the fa c to r sco res fo r th is co m p o n e n t are g reater than zero.
T w o co m m u n es co lo red in y e llo w sh o w a facto r score v e ry clo se to zero fo r th e first com ponent. T h is is Y ên N am (-0 .0 1 7 6 0 ) and M ộc N am (0 .0 6 4 2 5 ). T h e y rep resen t th e m odal situation o f th e đ istrict as p er no n-farm incom e.
T he rem a in in g co m m u n es are re p re se n te d in co ld colors. T h e re is rela tiv e ly less h o u seh o ld s in th e se c o m m u n e s e am in g n o n -farm in g revenue th a t in th e rest o f th e d istrict. O n e can observe th a t th e se lovv n o n -farm in g incom e com m un es a re m o stly located in th e Southern part o f the d isư ic t.
“A g ric u ltu re ằt
T h e m a p p in g o f resu lts for th is d im en sio n show s th at in tensive ag ric u ltu re te n d s to
70
p.v.
C u et a i / V N U Ịournal o f Science, Earth Sciences 25 (2009) 65-75c o n c e n tra te in the n o rth -cen tral p a rt o f the d is tric t. T h ere are seven co m m u n es vvith factor sc o re s su p erio r to 0 for the seco n d com po nent (T iê n H iệp, Y ên B ắc, T rác V ăn, C h âu G iang, T iê n N g o ại, T iên N ội an d Y ên N am ). T h ese are th e co m m u n es vvhere th e a g ric u ltu ra l sector is
th e m o st v igo rou s in th e d istrict based o n the rice p ad d les su rfaces and th e p ercen tag e o f the land rela tiv e ly ded icated to ag riculture. O n the o p p o site , co m m u n es vvith a less prom inent a g ricu ltu ral secto r (as co m p ared to th e d istric t’s averag e situ atio n ) are located a t th e periphery.
3.2. C a se stu d y 2: D e v e lo p m e n í o f the te rtia ry se c ío r / a g ric u líu re d im en sio n s
Table 5. Description o f variables used for case study 2
Variable Labcl (english) Description
@06Ho_XD
@06Ho_CNTTCN
@06Ho_TN
@06Ho_Van_tai
@06Ho_Dvu
@06DNN_Bqho Agriarea*________
C onstructionhh In du stry hh T ra d e h h T ran sp o rth h Service__hh agri_land_per_hh agri area^K
Number o f househoỉd involved in the construction sector.
Number o f household invoỉved in the industrial or small industry sector.
Number o f household involved in trading.
Number o f household involved in transportation.
Number o f household involved in the Service sector Surface o f agricultural land per household [m2]
Percentage o f total area dedicated to agriculture [ha]
T h e correlation m atrix c o m p u ted by SPSS show th e co rrelatio n s betw een the v ariab les used for th is c a se is rep roduced as sh ow n o n T able 6 below :
Table 6. Correlation mtrix
Induslry_hh Construction_hh Trade_hh Transport_hh Service hh agri land per hh agri arca pc
Industry_hh 1.000 .105 .374 .375 .254 -.472 -.450
Conslruction_hh .105 1.000 .317 .673 .608 -.247 .069
T radehh .374 .317 1.000 .680 1563 -.442 -.309
Transporthh .375 .673 .680 1.000 .713 -.343 -.100
Servicehh .254 .608 .563 .713 1.000 -.386 .072
agri_land_per_hh -.472 -.247 -.442 -.343 -.386 1.000 .531
agri_arca_pc -.450 .069 -.309 -.100 .072 .531 1.000
(B) Determinant = .033
O n e can alread y see from th e m atrix th at the v a ria b le s In d u stry _ h h , C on struction _h h, T ra n s p o rt hh and S e r v ic e J ìh rela te to each o th e r an d te n d to “clu ster” to stru ctu re a distinct co m p o n e n t. T h e calcu la ted o v erall K aiser’s m e a su re o f sam p lin g ad eq u a cy (K M O ) w as 0.7 0 3 w h ich is qu ite accep tab le in the c o n d itio n s o f th e study. T h e tw o first principal c o m p o n e n ts rep resen t m o re th a n 71% o f the v a ria n c e o f th e d atase t (4 7 .7 6 % and 23%
re la tiv e ly ) as sh ow n on T a b le 7.
Tabỉe 7. Quantity o f information represented by principle componenls
Component 1
2 3
Initial Eigenvalues Total % o f Variance 3.344 47.765
1.639 23.412 .597 8.532
Cumulative % 47.765 71.176 79.708
W e h av e then ex tra cted tw o com ponents u sing th e va rim a x rotation m ethod. T he Table 8 rep ro d u ced b elo w d isp lay s th e “ loadings" o f th e v aria b les on eac h com ponent.
p .v . Cu el aỉ. / V N U lournal o f Science, Earlh Sciences 25 <2009) 65-75 71
Table 8. Loading o f vatriables on each component.
Component
1 2
T ran sp o rth h .882 -.245
Service hh .882 -.108
Construction hh .829 .059
Trade hh .627 -.493
agri area_pc .161 .868
a g rija n d _ p e rh h -.295 .757
Industry hh .195 -.743
T h e v aria b les rc la te d to econ om ic activities such as tran sp o rt, c o n stru ctio n , trad e and other scrvices secto rs are group cd in the íìrst co m po nent. T his d im en sio n obviously rep resen ts th e level o f im p o rtan ce o f th e tertiary secto r in th e D uy T ien d is tric t’s econom y. A lso knovvn as th e Service in du stry o r Service sector w h ich d o es no t in v o lv e th e ex tractio n o f reso u rces n o r th e ir tran sfo rm a tio n b u t is based on the p ro v isio n o f se rv ic e s to bu sin esses as w ell as final co n su m ers01. T h e rem aining v aria b les are all stro n g ly related to th e second co m p o n en t w h ich te n d s to ag g reg ate the variab les th at directly rela te to farm in g o r show a stro n g in v erse re la tio n sh ip w ith it. T his is the case fo r th e n u m b e r o f fam ily in vo lved in sm all industries: o n e can a s s u m e th a t th e re is a clear inverse rela tio n sh ip betvveen th is v ariab le and th e im p o rtan ce o f fa rm in g activ ities in th e local econom y.
F a cto r s c o r e s m a p p in g
A s fo r th e íìrst c a se stud y p resen ted , the individual fa c to r sc o re s h av e b een ex tra cted for both c o m p o n e n ts (d e v e lo p m e n t o f tertiary secto r a n d a g ric u ltu re ) an d m ap p ed to sh o w the spatial d istrib u tio n o f the sco res. T h e m ap s are pro vid ed in th e A p p e n d ix B and com m en ted here in d etails. T h e m ap p in g also displays the estim ated q u a n tity o f w aste g en era ted in each com m une.
(l) Source: Insee (ỉnstitut nalional de la slalistique et des éludes économiques), France.
“A g ric u ltu re 'ẽ
T h e seco n d co m p o n en t show s a sim ilar ten d en cy in th e s c o re s ’ d istrib u tio n then th e one ob served fo r th e first case stu dy p resen ted (N o n -farm in g incom e vs. A g ricu ltu re). Som e d iíĩe re n c e s are n o ticeab le in the in ten sity o f ag ricu ltural activ ity . T h ese d iíĩe re n c e s are p artly d u e to th e fact th at th e q u a n tiỉe m eth o d p rop osed by A rcM ap w as used th is tim e as the c lassiíica tio n m e th o d to create th e gradu ated co lo r sy m b o lo g y instead o f th e n a tu r a ỉ breaks.
H ow ever, im p o rtan t part o f this d issim ilarity can m o reo v er b e ex p lain cd by inh eren t factors found in th e d ata such as p o p u la tio n den sity, and by th e ch o ice o f variables. F o r instance, T iên N g oại co m m u n e is gettin g th e h ig h est sco re m ainly b ecau se it has th e lovvest po pulation d c n sity o f th e d istrict an d th u s the hig hest su rface agricultu ral su rface p er ho usehold. O th e r ex am p les are Y ên B ắc and C h âu G ian g w h ich vvere shovving p rev io u sly th e h ig hest scores on th e ag ric u ltu re d im en sio n . T h is case stu d y a lso p o in ts o u t th at, w hen co n sid erin g a s w ell the ro le th a t th e sm all industries are p la y in g in th e se co m m u n ities, the agricultural s e c to r app cars less im p o rta n t than expected a t firs t glan ce. T h ese p ie c e s o f in íò rm atio n w e re n o t taken into a c c o u n t in the first case study.
"D evelo p m en t o f th e te rtia ry s e c to r ”
O n th e o th e r h an d , the m a p p in g o f íìrst co m p o n en t’s sco res sp atially illu strates and co m p ares th e d e v e lo p m e n t o f th e te rtia ry sector in D uy T ien . T he h ig h e st sco res cen tralize around b oth sm all to w n s esp ecially T T H oà M ạc. It is re a so n a b le to th in k th a t th e p ro x im ity o f an urban c e n te r is a facto r in th e g ro w th o f business activ ities. T h is co n sid era tio n seem s to have a greater im p act on th e d ev elo p m en t o f th e tertiary s e c to r th a t th e p ro x im ity o f a m ạịo r road.
72
p.v.
Cu et aỉ. / V N U Ịournaỉ o f Science, Earth Sciences 25 (2009) 65-753.3. C a s e s tu d y 3: N o n -fa r m in g in c o m e / B u iỉt-u p z o n e s e x p a n s io n d im e n s io n s
S ix v aria b les used to p e rfo rm th e p rin cip al c o m p o n e n t a n a ly sis o n th is c a se stu d y arc d escrib ed on T a b le 9 T ab le 9.
Table 9. Description o f variables used for case study 3 Variabỉe
T L ta n g
@06Ho_CNXD res ident_area00_06 D c h u y e n d u n g
@06DNN_Bqho
@06Ng_NLNTS
Label (english) pop_growth IC in c o m e
resident_area00_06 public_servive_area agri_ỉand_per_pe A F A_pc_income
Description
Population grovvth rate in %.
N um ber o f household with m ajor income from industry or construction.
Variation in residential suríace from 2000 to 2006 [%].
Land area used for pubỉic infrastructure (e.g. roads) [ha].
Suríace o f agricultural land per resident [m ].
Proportion o f household w hose mạịor income is from agriculture, forestry o r aquiculture. [%]_______________
T h e prin cipal c o m p o n e n t an a ly sis p e ríò rm e d h ad ex tra c te d m o re th a n 7 2 % o f th e v aria n ce o f th e d atase t a s re p o rte d b y S P S S in th e T a b le 10 p resen ted here.
Table 10. Quantity o f information extracted by each component.
Component Initial Eigenvalues
Total % o fV a ria n c e Cumulative %
1 2.751 45.845 45.845
2 1.585 26.410 72.256
3 .769 12.824 85.079
4 .557 9.284 94.363
5 .192 3.197 97.560
6 .146 2.440 100.000
U sin g y e t ag ain th e v a r im a x ro tatio n m eth o d , tw o c o m p o n e n ts vvere ex tra c te d by S P SS . T h e ta b le b elo w d is p la y s th e “ lo a d in g s”
o f th e ir resp ectiv ely re la te d v a ria b le s o n each c o m p o n en t.
Tabỉe 11. Loading o f variable on com ponents 1 and 2 in case study 3
Component
1 2
IC income .923 -.044
AFA_pc_income -.862 -.254
pub 1 ic s e rv i v e a r e a .821 -.165
pop_growth .075 .766
resident_area00_06 agri land per pe
-.223 -.567
.725 -.695
It a p p e a rs fro m th e ro tated com po nent m a trix th a t th e íìrs t d im e n sio n e x ừ a c te d reílects th e “ n o n -fa rm in g in co m e” c o n c e n tra tio n o f the D uy T ie n c o m m u n e s as in th e íirst case study p re v io u sly d is c u s s e d in d etails in this d o c u m e n t. T h e se c o n d co m p o n en t ex tra cted for th is c a s e h o ld s th e v a ria b le s related to d e m o g ra p h ic (p o p u la tio n g ro w th ) an d land use c h a n g e (p o s itiv e v aria tio n in residential area a n d n e g a tiv e v a ria tio n in a g ric u ltu ra l land area p er p erso n s). T h is d im en sio n rep resen ts the e x te n sio n o f th e b u ilt-u p a re a s i.e. th e land c o v e re d b y b u ild in g s and o th er m an-m ade stru c tu re s an d a c tiv itie s (2).
F a c to r s c o r e s m a p p in g
T h e m a p s a re p ro v id ed in th e A pp en dix
c.
It is in te re s tin g to h av e a d eep e r look a t the m a p p in g o f th e fa c to r sc o re s o f th e second c o m p o n e n ts (b u ilt-u p ex p an sio n ). T h e m ost n o tic e a b le b u ilt-u p e x p a n s io n d o e sn ’t necessary h a p p e n s o n ly o n th e o u ts k irts o f th e tw o to w n s as o n e c o u ld n o rm a lly ex p ect. S urprisingly, h ig h sc o re s are p re s e n t in so m e o f th e m o sl OÍT- ce n te re d c o m m u n e s su ch a s T iên P h o n g and
t2) As per the land cover categories proposed by the GTOS programme (http://www.fao.org/gtos/).
p .v . Cu et a i / V N U ìo u m a ì o ỊSãence, Earth Sciences 25 (2009) 65-75 73
Đọi Sơn. F o r T iên P hong, th is situ a tio n can be ex p lain ed tw o cau ses c o m b in e d . T h is co m m u nity p o ssesses th e g re a te s t p o p u la tio n grovvth o f the đ istrict. T h is in c re a se in po pulation puts p ressu re on th e re sid e n tia l area vvhich has increased by 6 0 % betvveen y e a rs 2000 and 2 0 0 6 vvhile the a v e ra g e v a ria tio n for th e d istrict w a s on ly 44 % . F u rth e rm o re , it is a fairly sm all co m m u n e; c o n s e q u e n tly it has initially a ra th e r sm all s u ría c e o f land u sed for ag riculture. S im ilar c o n sid e ra tio n s reg ard in g th e strong p op u latio n grovvth (1 .1 7 % ) an d an aggressiv e in crease in th e re sid e n tia l a re a (7 0 % ) can be ap p lied to th e Đọi S ơ n c o m m u n e .
4. C o n c lu sio n
By ex clu d in g T T Đ ồ n g V ă n a n d T T H oà M ạc, the tvvo sm all tovvns o f d istric ts, th e m edian m o n th ly q u an tity o f w a s te g e n e ra te d in th e rural co m m u n es is 9 k ilo g ra m s p e r p erso n s vvhile th e av erag e q u a n tity is 11 kg. A lm o st 60% o f th e co m m u n es (4 /7 ) w h e re th e fa c to r scores fo r th e “N o n -fa rm in g in c o m e”
dim ension are p o sitiv e sh o w a w a s te q u a n tity g reater th an th e average.
Hovvever, o n e c o m m u n e , n a m e ly Y ẽn B ắc, has a fairly high fa c to r sc o re (0 .8 5 3 0 ) fo r th is dim ension b u t g en era tes a ra th e r lo w q u a n tity o f w aste. It is in terestin g to n o te th a t th is com m une a lso sh o w s th e se c o n d h ig h e st facto r score for th e o th er d im en sio n (a g ric u ltu re ). O n e possible ex p lan atio n is th at th e re s id e n ts o f th is rural co m m u n e are also m ig ra n t w orkers betvveen th e cro p s. T h e y c a n s p e n d fe w m o n th s outside th e ir v illag e eac h y e a r w o rk in g in th e co n stru ctio n secto r in th e su rro u n d in g to w n s.
T he ad d itio n al rev en u e s e a m e d fro m th is seasonal w o rk w o u ld e x p la in w h y th is com m une b e a rs a h igh s c o re on th e “ n o n -farm incom e” co m p o n en t. O n th e o th e r h an d , th e fact that th e se in h ab itan ts are tc m p o ra ry liv in g aw ay from th e ir v illag e co u ld e x p la in th e ab o v e
a v e ra g e w a ste q u a n tity g en erated in the c o m m u n e.
M ộ c N a m c o m m u n e sh o w s an a p p re c ia tiv e ly a v e ra g e sco re fo r co m p o n en t
“N o n -fa rm in g in c o m e ” an d a v ery n eg ativ e sc o re o n “ a g ric u ltu re ” its w aste g e n e ra tio n is tw ic e th e a v e ra g e (2 2 kg). T h e c ra fì sector (p a rtic u la rly h a n d c ra fte d dye w o rk s) is w ell- d e v e lo p e d in th is oíT -centered co m m u n e. T his ty p ic a l a c tiv ity c o u ld ex p lain th e m o re than e x p e c te d w a ste g e n e ra tio n o f th e co m m u n e.
T h e p ro x im ity o f a m a jo r road a p p e a rs to h a v e a p o sitiv e im p a c t o n th e level o f no n-farm in co m e a t le ast fo r c o m m u n e s located a lo n g by th e m a in N o rth -S o u th an d W est-E ast a x is roads.
H o w e v er, th e re is n o o b v io u s relatio n w ith th e e x iste n c e o f a m a in ro ad Crossing a co m m u n e a n d th e w a ste q u a n tity g en era ted locally.
In re tro sp e c t, it is ap p a re n t th a t b ased th e in ío rm a tio n e x tra c te d fro m th e D uy T ie n data, th e q u a n tity o f w a s te g en erated in a g iven lo c ality is m o s tly d e te rm in e d b y its ủ in c tio n . T h e sm all to w n s o f th e d istrict c a rry o u t c o m m e rc ia l an d in d u stria l activ ities th a t a re not p re se n t (o r o n ly a t a m u c h less in ten se lev el) in th e ru ral c o m m u n e s . A s a c o n seq u en c e o f the ro le th a t th e tw o tovvns p lay as tra d e cen ters, th e se lo c a liz e d a c tiv itie s g en erate m o re w aste th a n a n y o th e r a c tiv itie s o rig in atin g from a n y w h e re in th e re s t o f th e district.
M in im a lly , th e re m u s t be m o re c a s e s than v a ria b le s to p e rfo rm a P C A . M a n y au th o rs m e n tio n th a t fa c to r a n a ly sis is in a p p ro p riate w h e n s a m p le siz e is b elo w 50. S o m e arb itra ry
"ru le s o f th u m b " a ls o e x ist and a re w id e ly used in p ra c tic e to c a lc u la te th e m in im u m n u m b e r o f ca se s re q u ire d . F o r in stan ce, a c c o rd in g to B ry a n t a n d Y a m o ld (1 9 9 5 ), th e n u m b e r o f c a s e s sh o u ld b e a t least 5 tim es th e n u m b e r o f v a ria b le s e n te re d in th e an aly sis. I t’s esse n tia l to m e n tio n th a t th e P C A repo rted h e re d o e s n ’t c o m p ly w ith th is ru le: on ly 2 1 c a s e s w ere a v a ila b le w h ile fiv e to seven v a ria b le s w ere in c lu d e d in th e m o d e l.
74 p.v. Cu et aì. / VNU Ịoumaỉ of Science, Earth Sciences 25 (2009) 65-75
Appendix A - Maps o f Factor Scores for Non-farming income - Agricuỉture
Appendix B - Maps o f the Factor Scores for Tertiary Sector - Agrìculíure
p.v.
Cu et al. / V N U Ịoum al o f Science, Earth Sciences 25 (2009) 65-75 75Appendix
c
- Maps o f the Factor Scores for Non-farming income - Buiỉt-up expansionF a c t o r S c o r e s f o r c o m p o n e n t "T e rtla ry S e c to r "
D u y T l « n d l s t r l c t , H a N a m p r o v t n c *
O u a n tlty o f w M l t
F a c to r S c o r e s f o r c o m p o n e n t " A g ríc u R u r* S « cto r"
D u y T t a n d l s t r t c t . H a N a m p r o v t n c *
R e íe re n c e s
[1] I. Agilent Technologies, Principal Components Anaỉysis, sip support@aeilent.com. 2005.
[2] Aurobindo Ogra. Logistics Management and Spatial Planning fo r Soỉid Waste Management System using Geographic Informaíion System Map Asia, 2003.
[3] Christian Zurbrtigg, Urban Soỉid ỈVaste Management in Low-Income Countries o f Asia How to Cope wiíh the Garbage Crisis, Urban Solid Waste Management Review Session, Durban, South Aĩrica, November, 2002.
[4] Đào Thắm,
về
đâu rác thải sinh hoạt nông thồn?http://www.baohungyen.vn/content/viewer.asp7a
=6848&z=63, 2007.
[5] Jan Pcter Lesschen, Peter H. Verburg, et al, Statisticaỉ methods fo r anaiysing the spatial dimension o Ị changes in ỉand use and /arming systems, The International Livestock Research Institute, Nairobi, Kenya & LUCC Focus 3 Office, Wageningen University, the Netherlands, 2005.
[6] M. McAdams, A. Demirci, The use o f principỉe componení anaỉysis ìn data reduction fo r GIS A naỉysis o f w aíer quality data, Voliune, DOI:, 2006.
[7] Marketing Dept. SPSS Inc., SPSS (Statistical Producers for Social Science), SPSS software and manuaỉ, Chicago, Illinois, USA, 2ỈXX).
[8]
