Translation 5099
Transcription
Translation 5099
ISSN 0704-3716 Canadian Translation of Fisheries and Aquatic Sciences No. 5099 A BASIC program to analyze the polymodal frequency distribution into normal distributions T. Akamine Original title: Polymodal-na Dosu Bunpu o Seiki Bunpu e Bunkai suru Basic Purgguramu In: Bull. Jpn. Sea Reg. Fish. Res. Lab. (33): 163-166, 1982 Original language: Japanese' Available from: Canada Institute for Scientific and Technical Information National Research Council Ottawa, Ontario, Canada R1A 0S2 1984 6 typescript pages , V II lee Secretary State Secrétariat d'État ' of MULTI LINGUAL SERVICES DIVISION — DIVISION DES SERVICES MULTILINGUES BUREAU DES TRADUCTIONS TRANSLATION BUREAU LIBRARY IDENTIFICATION — FICHE SIGNALÉTIQUE Translated from - Traduction de TrA5 5.0q Into - En English Japanese Author - Auteur Tatsuro AKAMINE Title in English or French - Titre anglais ou français A BASIC Program to Analyze the Polymodal Frequency Distribution into Normal Distributions Title in foreign language (Transliterate foreign characters) Titre en langue étrangère (Transcrire en caractères romains) Polymodal-na Dosu Bunpu o Seiki Bunpu e Bunkai suru Basic Puroguramu Reference in foreign language (Name of book or publication) in full, transliterate foreign characters. Référence en langue étrangère (Nom du livre ou publication), au complet, transcrire en caractères romains. Nihonkai-ku Suisan Kenkyusho Hokoku Reference in English or French - Référence en anglais ou français Bulletin of the Japan Sea Regional Fisheries Research Laboratory Publisher - Editeur Page Numbers in original Numéros des pages dans l'original DATE OF PUBLICATION DATE DE PUBLICATION Year Année Place of Publication Lieu de publication 1982 Requesting Department Volume 163-166 Issue No. Numéro Number of typed pages Nombre de pages dactylographiées 33 6 Ministère-Client Fi.sheries and Oceans Translation Bureau No. Notre dossier no Branch or Division Direction ou Division Scientific Information & Publication Translation (Initials) Traducteur (Initiales) Person requesting Demandé par A.T. Reid SE P 2 0 1984 Your Number Votre dossier no Date of Request Date de la demande SECS-111 (81/01) 1655823 September 6,, 1984 -- PS I* Secretary of State Secrétariat d'État MULTILINGUAL SERVICES DIVISION DIVISION DES SERVICES MULTILINGUES — BUREAU DES TRADUCTIONS TRANSLATION BUREAU Scientific Information and Publication Fisheries and Oceans 1355823 Ottawa Translator (Initials) — Traducteur (Initiales) Language — Langue Bureau No.—No du bureau City — Ville Division/Branch — Division/Direction Department — Ministère Clienes No.—N° du client SEP 2 0 1984 PS Japanese A BASIC Program to Analyze the Polymodal Frequency Distribution into Normal Distributions Tatsuro AKAMINE 1) Abstract The algorithm of this program is an iteration method which consists of the linear approximation and the method of least—squares. Variables are modified one after another. This program can be run on a micro computer, although it takes a longer time in comparison with larger computers. UNEDITED TRANSLAMIA TRADIJUiON le:1N RsvTir:E infzrrnai. irin zetikcnt Analyzing the polymodal frequency distribution into normal distribution is a fundamental method used when for instance dealing with animal species the age character of which is not known. Although several large computer programs have already been presented, researchers unable to use the above depend on such methods as that of Tanaka (1956). This basic program has been conceived for the presently marketed small computers of less than e100.000, and a 2K bite memory is sufficient. The calculating method which follows Shimadzu (1980) is that of linear approximation and of the least squares. The parameters are moved one at a time so as to minimize the memory. The fact that it takes time is a disadvantage but one has no choice but to take into consideration the performance of the small computers in question. 1) Japan SeaRegional Fisheries Research Laboratory, Suido-cho, Niigata 951, Japan. SEC 5-25 (Rev. 82/11) CanadI3 - 2 Method of calculating Let us call F the given frequency distribution, m the number of classes, h the width of the classes, a the smallest class value and b the greatest value. denoted by n. The number of normal distributions to be analyzed is The sought formulai, the residual functions d 2 are then: f =jKi. N (pi, 1 /27ra N x=. ee l j (x— p) 2 1 2a2 .dx2=E (F—f) 2 —Cg) Fi5, b = a + (m —1) h a h X 1 El smuyey] Fig. 1. Illustration of variables. The linear approximation o 2-1-'7 j r af gives: j .1_2/Jai 1aR- i 4K' +a-af --, i l's 2ai 1 • :if(i :.=_N (pi, ci, x) af =Ki • N (pi, ai, x) • e— ili 2 1 p' . cc i Cl • IV (pi, ui,x) (X — pi) 2 —ai 2 cri3 2 In order to minimize d , we make the parameters change in the order. If we indicate by a the parameter made to change we have: ace - Jo- - 3 - By substituting in d, — (f+4.1.)} '-= (d.— _iit4a)2 When using here the least square method, we obtain: ee —o ala o .(cr.M1 This is the corrected amount we are seeking. Example of application Using a MB 6880 microcomputer, we analyzed the body length composition of porgy by Tanaka (1956). d 2 is shown in Graph 2. The state of convergence of The results are shown in Table 1. in Fig. 3 is obtained by plotting X-Y. average of about 8 min for one time. The Graph With the MB 6880, it took an When finishing in 10 times, it takes about 80 min. Discussion With the present program, one must first put in the number of normal distributions and the initial values. As good initial values are put in data with clear modes, less time is needed. When to the contrary putting in bad initial values and data with no clear modes, vibrations and dispersion will frequently occur and there is a possibility of entering a stop point. This program is used to achieve objectivity and one must give plenty of consideration to input and output. 4 Improvement points of this program 1) improving the input, output 2) improving the loop escape judgement. At the close of this article, I wish to thank Fumihiko KATO, researcher responsible at the Resource Department of the Japanese Sea Regional Fisheries Laboratory, for drawing the chart based on the plotting of X-Y and for giving us much advice. e, 1. Number of Iterations Mqr* d'offliute Fig. 2. Convergence of d. nag FORK LENGTH 0113M ..e401,#R4E.Te. e") rDitt.:EK5145 t toeiontle IN CM 1'1 OD BASIC 7' e - Fig. S. Fork length frequency curve of porgy. The histogram shows the observed frequenXcy, lines normal curves and the sum of each normal curves fitted by this BASIC program. .41 --e 5 ••■• 2 ,ithiz umnoreR ON 1 Table 1. Comparison of the results from two methods. 2 1 3 4 5 cP K5000400030001000500 initial 27 15.5 20 24 value ir 11 1.5 1.5 1.5 758923 1 )12' 1 K 5676 4331 2703 798 492 value after 29 p 11.0215.2719.8223.4626.66 iterations a 0.8231.1621.4781.2021.4756314 TANAKA'S graphical method K 5627 4485 2630 840 461 10.9915.2619.8423.5026.82 0.8 1.2 1.4 1.2 1.4 19192 Correspondence of variables 14 :numberofnormaldistributions A : Minimum class mark H :class width M number of classes I)(1,J) K1 81 D(2,1) :II; I:11:d. I)(3,J) 112 :d. :fri • F (X) 9.0 J- 9 Program list F., G1: 0L A2 : da e 05' — P 7‘11f4U) (DATA : an exampkof porgy) REM POLVMODAL REM :0019 READ N.A.M,H PRINT N= • 10 PRINT '4= . 1A50 60 PRINT '11='10 70 PRINT • H • 114 80 DIM 013,01,F(M) 90 STOP 100 FOR 1=1 TO M 110 READ FM:PRINT 'F( . :11')= . tFt1/ • 120 NEXT I 10 20 30 40 130 SIOP 140 FOR J=1 TO N 150 FOR 1=1 TO 3 160 READ DCI.J/tPRINT 170 NEXT 1s56XT J 180 STOP 190 READ 091PRINT '09= . 109 200 STOP 210 REM 141) 220 D7=1 230 FOR 1=1 TO 3 240 FOR J=1 TO N 250 02=0: 5 1=0182=0 260 FOR 5 =1 10 14 270 0=14e(K-1I=H 280 F1=0 290 FOR L=1 TO N 300 P1=011.1):P2=X-0(2,L)1P3=0(3,1) • 310 91=91e1 1 ■ .39C^42:73«E.St-.5=FZ , P2,P3/F3) 320 NEXT L 130 01=F(10-F1 340 01=13(1,J)s02=X-0(2,J)t03=0(3,J) 350 01=.398942/03.ExPI-.51.02+02/03/031 360 ON 1 GOTO 390.370,380 370 GI=C11..G1=02/03/031G0T0 390 380 01=01«01..(02•02-03=03)/03/03/03 390 02=02.01=01 400 S1=014E0001 410 S2=52.61=01 420 NEXT K 430 14 2= 5 1/S2 440 0i1.J>=0(1,J).02 450 PRINT 460 PRINT 137='107 470 PRINT • 1='11, • ,.1=';J 480 PRINT '02='032.'A2='1A2 490 NEXT JsNEXT I 500 07=07+1 510 IF 07<=09 GOTO 230 520 REM 51 5 9,9 530 PRINT 540 FOR J=1 TO N 550 FOR 1=1 10 3 560 PRINT 13I . ;11 . ..001=';011,J1 570 NEXT IINEXT J 500 ENÛ 1000 DATA 5.7.5,29.1 1010 DATA 7.79.509.2240.2341.623,476.1230.1439 1020 DATA 921.448,512,719.673.445.341.310.228 1030 DATA 168,140.114.64,22,0,2.2.0.0.1 1040 DATA 5000.11.1.4000,15.5.1,3000,20.1.5 1050 0010 1000,24.1.5,500.27.1.5 1060 DATA 30 6 Bibliography Yasuhiko SHIMAZU Taicho Sosei kara Nenrei Sosei o Suitei suru Hoho (Method of evaluating the age composition from the body length composition) Showa 54 nendo Gyogyo Shigen Kenkyu Kaigi, Nishi Nihon Teigyo Bukai Kaigi Hokoku (The 1979 report from the Western Japan Bottom Fish Branch Conference, Fisheries Resources Research Council): Shoichi TANAKA (1956) 36-48. Polymodal-na Dosu Bunpu no Hitotsu no Toriatsukaikata oyobi sono Kidai Taicho Sosei Bunseki e no Oyo (One way of handling polymodal frequency distribution and application to the analysis of porgy body length composition) Tokaisuiken Hokoku (abbreviated title; from the Eastern Fisheries Laboratory) (14): 1-13. e eatVi1:(1979). f*R411. ■1a , r7ink1591 venef-ensequng-ir. miEl*ItAteeezegei: 36-48. Polymodal leltekiT - eop — ° (03i3ZIECRZY-Ice) eRtibegtn-N ape-A. *ei(lifulifef (14) :1-13. report