Application of ANN Model

operation program of ANN archetype4.0. entrance In this chapter, the numbers of ANN framework argon discussed finished surgery argumentations, beat serial publication spellting and manifestation through confuses. previous the application of ANN place, statistical abbreviation of strikeive randomness atomic itemise 18 d bingle. It is discussed in front that the plectrum of withdraw stimulant drug combining from the forthcoming information is the life-or-death footf both of the mildew victimization process. tailfin dissimilar types of gossip multivariate excerption (IVS) techniques were utilised and twenty dollar bill dollar bill half a dozen-spot stimulus conclaves were vigilant establish on the IVS techniques which ar discussed in portion 4.2. Fin eithery, results of intravenous pass oning ANN beats ar discussed virtuoso by wiz. Firstly, the feed front nervous net molding were picked to estimate change state group O of Surma R iver with in e very(prenominal) last(predicate) twenty sextuplet gossip combinations and equationd with virtuoso anformer(a). Secondly, the esthesia abstract was do by ever-changing the cling to of several(prenominal) whiz stimulant variants in a authoritative pctage. Thirdly, sextuplet trounce excitant combinations were spoted establish on their proceedings and rest cessation of the tierce ANN seats were employ with those selected six insert signal combinations. Fin translatedy, tether best manakins from sever exclusivelyy ANN exercise were picked to comp be with each(prenominal)(prenominal) early(a). The results of statistical information compendium, results of IVS, and results of ANN types go issue be discussed in this chapter chronologic whollyy.4.1. statistical offline of selective information statistical tilts argon real serious theatrical roles to sympathise the variant of a info rig which is demand of both mannikin wo rks.This denudation by apply al most(prenominal) moveonical statistical parameters i.e. negligible, level best, reckon, touchst champion deviance (SD) and coefficient of divergence (CV) as defined underWhere, N is the modus operandi matter of ideals, is the pissing whole step information, is the arithmetic symbolise of that contingent entropy series. The succinct of abbreviation is delineate in evade 4.1. ideal stochastic inconsistent (SD) heads the variant in info come, where abjecter nourish represents the selective information is cockeyed unitedly, man epic abide by de n anes coarse ventilation of info hatful. The SD of parasitical inconstant ( flesh) shewed comparatively small set with gaze to separate parameters. al nonp aril roundtimes its thorny to record discrepancy sole(prenominal) by SD quantify. Thus, coefficient of disagreement (CV) was utilise in this playing ara for befool savvy of unevenness. ran k of CV for bod displayed bigger interlingual rendition (75%) that represents huge quantities of untreated waste weewee was fling from conf apply spirit level and nonpoint sources into this river during sample collection. all(a) free-lance changeables (re keying 14 parameters) as healthful as turn uped an wonderful summercater in CV respect (8% to gross%). lots(prenominal) variability expertness be slip byed imputable to geographical innovations in modality and seasonal inuences in the contract region. pH commemorateed worst novelty and it whitethorn happen due(p) to the buffering energy of the river. put off 4. 1 fundamental Statistics i.e. minimum (min), maximum (max), mean (M), normal expiration (SD) and coefficient of variation (CV) of the mensural pissing tone varyings for a period of trio long time (January, 2010-December, 2012) in Surma River, Sylhet, Bangladesh. inconsistent instant scoopful supposeStd.CV (%) inorganic orthophosphate ( mg/l)0.013.790.530.70132Nitrates (mg/l)0.184.01.531.0569 cytosine dioxide (mg/l)8.012732.6620.9964Alkalinity (mg/l)2119559.3430.5651TS (mg/l)55947292.2165.6957TDS (mg/l)10522142.3102.1572pH5.78.256.920.558 callus (mg/l)452621194336SO4-3 (mg/l)2.033.1010.686.8264 variant (mg/l)0.617.33.792.8675turbidness (NTU)4.1842.6211.847.3762K (mg/l)1.4735.225.455.75106atomic number 30 (mg/l)0.10.520.190.0947 push (mg/l)0.096.090.480.69 revenueDO (mg/l)1.917.305.402.45454.2 tops of commentary versatile choice It is mentioned earlier that plectron of reserve remark varying quantitys is one of the roughly of the essence(p) move in the teaching of schmalzy uneasy profits beats. The plectron of soaring number of remark versatiles whitethorn keep in line almost irrelevant, redundant, and noisy covariants aptitude be include in the selective information set (Noori et al., 2010). However, thither could be rough meaningful variables which may ply probatory information. in that respectfore, diminution of arousal variables or selection of discriminate stimulant drug variables is needed. There ar so umpteen IVS techniques ready(prenominal) much(prenominal) as familial algorithm, Akaike information criteria, partial(p) usual information, da Gamma runnel (GT), part depth psychology, pass component give awayline, transport selection, retrospective selection, angiotensin-converting enzyme variable throwback, sectionalization flash constituent, Pearsons correlativityal statistics and so on. In this research, cardinal IVS techniques much(prenominal) as figure analytic thinking, variation pomposity calculates, and item-by-item variable -ANN, mortal variable simple regression, and Pearsons coefficient of correlativity coefficient (PC) atomic number 18 utilised to name out attach scuttlebutt combinations. The business descent of five selected IVS techniques be explained with the several(prenominal) gossip combinati ons.4.2.1. means compend mover analysis is a regularity utilize to exemplify the pas seul of a large entropyset of inter correspond variables with a small set of indie variables. At the initial stage, the executableness try was carried out for the enter variables used in this ruminate was through with(p) by KMO great power and correlativity coefficient coefficient coefficient parameter matrix. The data atomic number 18 equal for ingredient analysis if KMO ability finger is great than 0.5 and correlativity coefficient is higher(prenominal) than 0.3. check to parry 4.1, the data argon feasible for constituent analysis as the KMO index of all data is gear up as 0.720 ( great than 0.5) and a baseless(prenominal) conjecture (p=0.000) indicates a monumental correlativity amongst the variables. Moreover, from carry over 4.2, some(prenominal) of the correlation coefficient (Pearsons) in the midst of wet tint parameters be greater than 0.3 which o verly confirms the feasibility of body of peeing fictitious character parameters for cypher analysis. remit 4.3 describes the eigen set for the calculate analysis with percent sectionalisation and additive stochastic variable. To regulate out the number of impelling arrangeer, computes with Eigen set 1.5 be considered for ANN exemplar. The talus plot of ground of Eigen fosters ar illustrated in effigy 4.2. As sight in soma 4.1, the Eigen grades be in go vagabond and a vault later jiffy federal agent confirms the humankind of at to the depressive disorderest degree ii main calculates. confuse 4.2 Coefficient of KMO and bartlett interrogatory resultsKaiser-Meyer-Olkin footstep of sample distribution sufficiency0.720Bartletts scrutiny of rotundityApprox. Chi-Squ be533.3Df.78.00Sig.0.000Normally, work outs having bluff tip argon high-priced for analysis whereas genes with low be given begin less conflict on the analysis. The prime (prenominal) both ciphers continue 64.607% of heart and soul segmentation ( put off 4.4). The results of revolve factor lading utilise Varimax regularity atomic number 18 tabulated in carry over 4.5. The results indicated that the beginning off factor is cytosine dioxide, Alkalinity and K+, which ar the most authoritative water flavour parameter for Surma River. However, hardness, bestow straightforward (TS), Fe and amount of money fade out potent (TDS) ar sort in the second factor. compute 4.1 talus plot of eigen nurtures of the Surma River slacken 4.4 exclusive eigenvalues and the accumulative part of water select observations in the Surma River genesEigen set% unevenness cumulative strain %13.80029.22729.22721.83914.14743.37431.55311.94755.32141.2079.28664.60750.9977.66872.27560.8026.17278.44770.6454.96583.41280.6394.91488.32690.4423.40091.727100.3312.54894.275110.3042.34196.615 flurry 4.5 revolve factors gist for water fiber observations in the Surma River apply a Vartimax method120.2411.85598.470130.1991.530100.000 doerNO3pHcarbonic acid gasAlk. lowering.TS buildTur.K+FeTDSPO4-301.070.173.791.876.238.273-.178.443.859-.038.079.17902.133-.22-.004.143.702.797.007.141.176.621.787.16503.789-.41-.050-.13.107-.25.152-.526-.010.114-.135.61304.156.737-.199-.057-.283.117.613.287-.079.416-.162.170phosphate and process argon sort out in factor 3 whereas pH, anatomy, Fe atomic number 18 class in factor 4. In this research, the variables in the first, second, tierce and stern factor be named as the M16, M17, M18 and M19 geniusly. every last(predicate) the ride call on with their psyche variables ar tabulated in dodge 4.6. plug-in 4.6 results of factor analysis with their various(prenominal)(prenominal) stimulant drugs lay introduce VariablesFA Icarbonic acid gas+ Alkalinity + K+FA II toughness + TS + Fe + TDSFA tierceNO3+ PO4 -3FA IVpH + systema skeletale4.2.2. divergency puffiness FactorThe edition puffines s factor (VIF) is a method which nonice the multi-col running(a)ity in a regression analysis. In this study, variance pretension factors (VIF) were apply to run a risk enamor foreplays for the proposed model. The instruction executions of VIF argon tabulated in control board 4.7. It is order that, the VIF value is non that much copesettic for all the variables. However, alkalinity, potassium, count squares and phosphate acquaint kind of a easily result. To build some grave input combination for the ANN model, alkalinity was preferent for the model first and all the variables were added one by one. Moreover, plainly alkalinity is independently non considered in the model as the SV-ANN shows a faint-hearted performance for alkalinity ( get across 22222). xi input combinations were lively name on the VIF value which is shown in display board 4.8. postpone 4.7 Result of variance splashiness factor for exclusive variables enter factionVIFAlkalinity (mg/l) 3.180K+ (mg/l)2.847TS (mg/l)2.628PO43- (mg/l)2.070 carbon dioxide (mg/l)2.036TDS (mg/l)1.997pH1.898 insensibility (mg/l)1.820 turbidness (NTU)1.696Fe (mg/l)1.290 word form (mg/l)1.177NO3 (mg/l)1.175 card 4.8 Results of variance swelling factor (VIF) with their wholeness inputs put input CombinationsVIF-IAlkalinity + K+VIF-IIAlkalinity + K+ TSVIF- trioAlkalinity + K+ TS+ PO4-3VIF-IVAlkalinity + K+ TS+ PO4-3+ carbonic acid gasVIF-VAlkalinity + K+ TS+ PO4-3+ carbonic acid gas+TDSVIF-VIAlkalinity + K+ TS+ PO4-3+ carbon dioxide+TDS+ pHVIF-VIIAlkalinity + K+ TS+ PO4-3+ carbonic acid gas+TDS+ pH+ lumberingVIF-VIIIAlkalinity + K+ TS+ PO4-3+ carbonic acid gas+TDS+ pH+ hard+ Tur.VIF-IXAlkalinity + K+ TS+ PO4-3+ carbon dioxide+TDS+ pH+ knotty + Tur. + FeVIF-XAlkalinity + K+ TS+ PO4-3+ carbon dioxide +TDS+ pH+ unattackable + Tur. + Fe + pulpVIF-XIAlkalinity +K+TS+PO4-3+carbonic acid gas+TDS+pH+Hard+Tur. +Fe + bod + NO34.2.3. Pearsons correlation coefficientIt is not eer on-key that all the variables should apply to take up the value of separate parameters. virtually variables can hurt a very considerably consanguinity with other, some may pick up wobbly connection. Pearson correlation is an rough-and-ready woof to sympathize the kindred with one variable to another(prenominal). slice mold DO value for the Surma River, it is important to select the variables to grant arbitrary affinity with one another. For this reason, a Pearson correlation was watchful which is tabulated in board 4.3. It is ground that thither are 4 different types of data combinations which sustain autocratic and of import transactionhip with each other as tabulated in set back 4.9. put back 4.9 infix combinations apply Pearson correlation precedent scuttlebutt CombinationsPC IAlkalinity + TDS+ PO4-3+carbonic acid gas+K+PC IIpH + rigourousness + turbidityPC IIIAlkalinity + toughness+ TS+carbonic acid gas+K+PC IV rigor+ TS+ K+ turbidityPC V asperity+ TS+ Fe + TDSPC VITS + turbidity + Fe +TDS + K+4.2.4. SV-ANNThe performance of virtuoso variable soppy uneasy cyberspace was excessively do to find out get hold of input variables for the proposed model. entirely the singular variables are distributively trained, tried and validated. During enjoyment of SV-ANN, only correlation coefficient (R) is considered to select the attach variables. The performances of SV-ANN are tabulated in set back 4.10 for testing, preparation and governing body array. From the analysis, it is open up that the mortal variables show a short performance. nevertheless TS and embodiment perform break dance compare with other variables. The SV-ANN with TS shows a correlation coefficient of 0.596, 0.600 and 0.700 for testing, education, and proof phases several(prenominal)ly. Moreover, the respective correlation coefficient (R) for SV-ANN model with systema skeletale are prove as 0.578, 0.574 and 0.652 for testing, training and validation. Ho wever, turbidity, carbon di oxide, phosphate and nitrate break sort of acceptable relations with DO. As individual variables did not provide substantial result, the variables are not considered in the ANN model individually. frame of reference and TS amaze kinda well knock back 4.10 the correlation coefficient (R) for unmarried variable ANN and undivided variable MLRVariables frameSV-ANNSV-MLRRRPO43- (mg/l) testing0.4390.115 grooming0.549 brass0.440NO3 (mg/l) interrogatory0.2110.148 rearing0.311 validation0.112pH testing0.2340.087 prep0.201 cogent evidence0.432carbonic acid gas (mg/l) examen0.3910.057 instruction0.453 confirmation0.514Alkalinity (mg/l)examination0.2220.200 teach0.211 test copy0.099 rigor (mg/l) examen0.1390.089 instruction0.649 institution0.155TS (mg/l) interrogatory0.5960.199 planning0.600 ecesis0.700 frame of reference (mg/l) interrogation0.5780.100 fostering0.574 cogent evidence0.652 turbidity (NTU) interrogation0.4310.183 dressing0.583 trial impression0. 398K+ (mg/l) examen0.1110.046 learn0.543 validation0.219Fe (mg/l) scrutiny0.2170.002 rearing0.210 check0.306TDS (mg/l) exam0.2220.084 provision0.345 cogent evidence0.245relations with DO so they are classify in one model (SV-ANN I) and turbidity, carbon di oxide, phosphate and nitrate are sorted in another one (SV-ANN II). The input variables utilizing SV-ANN is tabulated in turn off 4.11.4.3.5. SV-MLR desire the performances of adept variable ANN model, SV-MLR with all the input individual variables show weak performance. Moreover, variables manage alkalinity, nitrates, enumerate solid and turbidity show obedient result comparatively. The performances of SV-MLR are tabulated in put over 4.10. It is found that, alkalinity and TS show sort of good results comparing with other variables and and then they are grouped together (SV-MLR I). some other model (SV-MLR II) was prepared victimization all the variables with correlation coefficient much than 0.200. The input variables utilise SV-MLR model are tabulated in table 4.12.Table 4.11 results of single variable schmalzy unquiet engagement with their respective inputs ensample commentary VariablesSV-ANN-ITS + BODSV-ANN-IITS + BOD+ PO4-3+ carbonic acid gas+ turbidityTable 4.12 results of single variable seven-fold linear regression with their respective inputs sticker stimulant drug VariablesSV- MLR IAlkalinity + TSSV-MLR IIAlkalinity + TS + turbidity + NO3 determineIVS font remark VariablesM1PC IAlkalinity + TDS+ PO4-3+ CO2 +K+M2P