A study on the ability of Support Vector Regression and Neural Networks to Forecast Basic Time Series Patterns

Size: px
Start display at page:

Download "A study on the ability of Support Vector Regression and Neural Networks to Forecast Basic Time Series Patterns"

Transcription

1 A study on the ablty of Support Vector Regresson and Neural Networks to Forecast Basc Tme Seres Patterns Sven F. Crone, Jose Guajardo 2, and Rchard Weber 2 Lancaster Unversty, Department of Management Scence, Lancaster LA 4YX, Lancaster, UK s.crone@lancaster.ac.uk 2 Unversty of Chle, Department of Industral Engneerng, Republca 70, Santago, Chle {jguajard,rweber}@d.uchle.cl Abstract. Recently, novel learnng algorthms such as Support Vector Regresson (SVR) and Neural Networks (NN) have receved ncreasng attenton n forecastng and tme seres predcton, offerng attractve theoretcal propertes and successful applcatons n several real world problem domans. Commonly, tme seres are composed of the combnaton of regular and rregular patterns such as trends and cycles, seasonal varatons, level shfts, outlers or pulses and structural breaks, among others. Conventonal parametrc statstcal methods are capable of forecastng a partcular combnaton of patterns through ex ante selecton of an adequate model form and specfc data preprocessng. Thus, the capablty of sem-parametrc methods from computatonal ntellgence to predct basc tme seres patterns wthout model selecton and preprocessng s of partcular relevance n evaluatng ther contrbuton to forecastng. Ths paper proposes an emprcal comparson between NN and SVR models usng radal bass functon (RBF) and lnear kernel functons, by analyzng ther predctve power on fve artfcal tme seres: statonary, addtve seasonalty, lnear trend, lnear trend wth addtve seasonalty, and lnear trend wth multplcatve seasonalty. Results obtaned show that RBF SVR models have problems n extrapolatng trends, whle NN and lnear SVR models wthout data preprocessng provde robust accuracy across all patterns and clearly outperform the commonly used RBF SVR on trended tme seres. Introducton Support Vector Regresson (SVR) and Artfcal Neural Networks (NN) have found ncreasng consderaton n forecastng theory, leadng to successful applcatons n tme seres and explanatory forecastng n varous domans, ncludng busness and management scence [, 2]. Methods form computatonal ntellgence promse

2 2 Sven F. Crone, Jose Guajardo and Rchard Weber attractve features to busness forecastng, beng data drven, sem-parametrc learnng machnes, permttng unversal approxmaton of arbtrary lnear or nonlnear functons from examples wthout a pror assumptons on the model structure, often outperformng conventonal statstcal approaches of ARIMA- or exponental smoothng- methods. Despte ther theoretcal capabltes, NN as SVR are not establshed forecastng methods n busness practce. Recently, substantal theoretcal crtcsm of NN has rased skeptcsm regardng ther ablty to forecast even smple tme seres patterns of seasonalty or trends wthout pror data preprocessng [3]. Whle all novel methods must ultmately be evaluated n an objectve experment usng a number of emprcal tme seres, adequate error measures and multple orgns of evaluaton [4], the fundamental questons to ther ablty to approxmate and generalze basc tme seres patterns must be evaluated beforehand. Tme seres can generally be characterzed by the combnaton of basc regular patterns: level, trend, season and resdual errors. For trend, a varety of lnear, progressve, degressve and regressve patterns are feasble. For seasonalty, an addtve or multplcatve combnaton wth level and trend further determnes the shape of the fnal tme seres. Consequently, we evaluate SVR and NN on a set of artfcally created tme seres derved from prevous publcatons. We evaluate the comparatve forecastng accuracy of each method to reflect ther ablty of learnng and forecastng fundamental tme seres patterns relevant to emprcal forecastng tasks. Ths paper s organzed as follows. Frst, we provde a bref ntroducton to SVR and NN n forecastng tme seres of observatons. Secton three presents the artfcally generated tme seres and the expermental desgn. Ths s followed by the expermental results and ther dscusson. Conclusons are gven n secton 4. 2 Modellng SVR and NN for Tme Seres Predcton 2. Support Vector Regresson We apply the common Support Vector Regresson (SVR) algorthm as proposed by Vapnk [5], whch uses an ε-nsenstve loss functon for predctve regresson problems. Ths functon allows a tolerance degree to errors not greater than ε. The descrpton s based on the termnology used n [6, 7]. Let {(x,y ),.., (x l,y l )}, where x є R n and y є R, be the tranng data ponts avalable to buld a regresson model. The SVR algorthm apples a transformaton functon Φ to the orgnal data ponts from the ntal Input Space, to a hgher-dmensonal Feature Space F. In ths new space, we construct a lnear model, whch corresponds to a non-lnear model n the orgnal space : When Φ s the dentty functon, the Feature Space s equvalent to the Input Space, and the model constructed s lnear n the orgnal space.

3 A study on the ablty of Support Vector Regresson and Neural Networks 3 Φ : R n F, w F f ( x) = w, Φ( x) + b The goal when usng the ε-nsenstve loss functon s to fnd a functon that fts current tranng data wth a devaton less or equal to ε, and at the same tme s as flat as possble. Ths means that one seeks for a small weght vector w; one way to do that s e.g. by mnmzng the quadratc norm of the vector w [6]. As ths problem could be nfeasble, slack varables ξ, ξ* are ntroduced to allow error levels greater than ε, arrvng to the formulaton proposed n [5]: l 2 * Mn w + C ( ξ + ξ ) 2 = s. t. y w, Φ( x ) b ε + ξ w, Φ( x ) + b y * ξ, ξ 0, =,2,..., l ε + ξ Ths s known as the prmal problem of the SVR algorthm. The objectve functon takes nto account generalzaton ablty and accuracy n the tranng set, and embodes the structural rsk mnmzaton prncple [8]. Parameter C measures the trade-off between generalzaton ablty and accuracy n the tranng data, and parameter ε defnes the degree of tolerance to errors. To solve the problem stated above, t s more convenent to represent the problem n ts dual form. For ths purpose, a Lagrange functon s constructed, and once applyng saddle pont condtons, t can be shown that the followng soluton s obtaned [8]: l * w = ( α α ) Φ( x ) l = * * f ( x) = ( α α ) K( x, x) + b = Here, α and α * are the dual varables, and the expresson K(x,x) represents the nner product between Φ(x ) and Φ(x), whch s known as the kernel functon [8]. The exstence of such a functon allows us to obtan a soluton for the orgnal regresson problem, wthout explctly consderng the transformaton Φ(x) appled to the data. In our experments we use radal bass functons (RBF) and lnear kernel functons. Lmted research has been conducted to nvestgate the ablty of SVR for predctng dfferent tme seres patterns. Experments performed by Hansen et. al [9] compare SVR performance wth 3 statstcal methods (e.g. ARIMA) on predctng 9 dfferent patterns present n real world tme seres. Among other patterns, they tred trends, seasonalty, cycles, and combnatons of them. Ther experments show SVR models outperformng the other methods on 8 of the 9 patterns; partcularly, they obtaned very good results usng SVR for extrapolatng lnear and non lnear trends. Guajardo et al. [0] compared SVR wth ARMAX models for predctng seasonal tme seres n a weekly sales forecastng doman for 5 dfferent products. Ther experments show that SVR were slghtly better than ARMAX models, succeedng n extrapolatng seasonal patterns (wthout trends) wth SVR.

4 4 Sven F. Crone, Jose Guajardo and Rchard Weber 2.2 Neural Networks Forecastng wth non-recurrent NN may encompass predcton of a dependent varable ŷ from lagged realzatons of the predctor varable y t n, or explanatory varables x of metrc, ordnal or nomnal scale as well as lagged realzatons thereof, xt, n. Therefore, NNs offer large degrees of freedom towards the forecastng desgn, permttng explanatory or causal forecastng through estmaton of a functonal relatonshp of the form yˆ = f ( x, x,..., x 2 z ), as well as general transfer functon models and smple tme seres predcton. Followng, we present a bref ntroducton to modellng NN for tme seres predcton; a general dscusson s gven n [, 2]. Forecastng tme seres wth NN s generally based on modellng the network n analogy to a non-lnear autoregressve AR(p) model [2, 3]. At a pont n tme t, a one-step ahead forecast y ˆ t+ s computed usng p=n observatons yt, yt, K, yt n+ from n precedng ponts n tme t, t-, t-2,, t-n+, wth n denotng the number of nput unts of the NN. Ths models a tme seres predcton as of ( y, y y ) ˆ = f,..., t + t t t n+ y. The archtecture of a feed-forward Multlayer Perceptron (MLP), a well researched NN paradgm, of arbtrary topology s dsplayed n fgure y ˆt+ y ˆt+ 2 y ˆt+ h Fg.. Autoregressve MLP applcaton to tme seres forecastng wth a MLP of arbtrary topology, usng n nput neurons for observatons n t, t-, t-2,, t-n-, m hdden unts, h output unts for tme perods t+, t+2,, t+h and a two layers of tranable weghts. The bas s dsplayed wthn the unts. Data s presented to the MLP as a sldng wndow over the tme seres observatons. The task of the MLP s to model the underlyng generator of the data durng tranng, so that a vald forecast s made when the traned NN s subsequently presented wth a new nput vector value.

5 A study on the ablty of Support Vector Regresson and Neural Networks 5 The network paradgm of MLP offers extensve degrees of freedom n modellng for predcton tasks. Structurng the degrees of freedom, each expert must decde upon the selecton and samplng of datasets, the degrees of data preprocessng, the statc archtectural propertes, the sgnal processng wthn nodes and the learnng algorthm n order to acheve the desgn goal, characterzed through the objectve functon or error functon. For a detaled dscusson of these ssues and the ablty of NN to forecast unvarate tme seres, the reader s referred to [2]. 3 Experments and Results 3. Descrpton of the Artfcal Tme Seres We evaluate a set of fve artfcal tme seres of monthly retal sales motvated from Pegel s orgnal classfcaton, later extended by Gardner to ncorporate degressve trends. Tme seres are composed of regular patterns of dfferent forms of lnear, progressve, degressve or regressve trends T, addtvely or multplcatvely combned wth seasonalty S, a constant level L and resdual nose E. In addton, emprcal tme seres are mpacted by rregular patterns such as level shfts and pulses, whch are dsregarded. To evaluate the ablty of dfferent computatonal ntellgence methods we create a set of benchmark tme seres for the most common regular tme seres patterns: lnear trend and dfferent forms of seasonalty. Consequently, we create ndvdual tme seres patterns and combne them accordngly, overlayng each wth addtve nose. No Trend (L) No Seasonalty (E) Addtve Seasonalty (S A ) Multplcatve Seasonalty (S M ) Lnear Trend (T L ) Fg. 2. Basc tme seres patterns of artfcal tme accordng to the Pegels- and Gardnerclassfcaton, combnng Level, Trend and Seasonalty wth a medum addtve nose level. In contrast to Pegel s classfcaton, a tme seres wth multplcatve seasonalty L+S M +E cannot dsplay an ncreasng seasonalty n the absence of level changes, t equals the pattern of addtve seasonalty and was consequently omtted from further analyss. Consequently, we create a set of fve tme seres ncludng a statonary tme seres L+E (E), seasonalty wthout trend L+S A +E (S A ), lnear trend L+T L +E (T L ), lnear trend wth addtve seasonalty L+T L +S A +E (T L S A ) and lnear trend wth multplcatve seasonalty dependng on the level of the tme seres L+T L *S M +E

6 6 Sven F. Crone, Jose Guajardo and Rchard Weber N applyng a medum level of nose σ 2 = 25. The orgnal tme seres data was taken from the experments of [3] and represent monthly retal sales. All tme seres consdered an addtve nose term to allow an estmaton of fnal forecastng accuracy n relatonshp to the orgnal nose level. Each tme seres conssts of 228 observatons. 2 (T L S M ). The resdual error term follows a Gaussan dstrbuton ( 0, σ ) 3.2 Expermental Desgn Ths research nvestgates whether the fve patterns descrbed above can be accurately predcted wth RBF SVR, Lnear SVR and NN models. For each seres, we defned a lag structure ncludng the 3 prevous observatons as attrbutes for predctng the next seres value (one perod ahead predcton); thus, a total of 25 data ponts reman to buld and parameterze models. Data was sequentally dvded nto tranng, valdaton and test sets usng 9, 48 and 48 observatons respectvely; tranng data s used to buld the model, valdaton data for parameter selecton purposes, and test data to evaluate the accuracy on a hold-out data set. All models are parameterzed usng only tranng and valdaton data, wthholdng all nformaton n the test set (also for scalng etc.) to assure vald ex ante testng. Data was transformed only by applyng lnear scalng nto a [-0.5, 0.5] nterval to avod saturaton effects, usng mnmum and maxmum values only from the tranng and valdaton data. No other preprocessng procedures such as deseasonalzaton or detrendng were carred out. As mentoned n secton 2.., SVR models requre settng of two parameters: C and ε. In addton, one needs to select an approprate kernel functon to carry out the transformaton to a hgher dmensonal feature space. The RBF kernel functon, whch s the kernel functon most wdely utlzed for regresson (see e.g. [6, 4, 5]), requres the defnton of an addtonal parameter σ. Our heurstc approach for RBF SVR parameter selecton can be summarzed as follows: - Frst, we determne startng values for the C and ε parameters on each tme seres by usng the emprcal rules proposed by Cherkassky and Ma [4], leadng to E {C= ; ε= }, S A {C= ; ε= }, T L {C= ; ε= }, T L S A {C=0.7464; ε= } and T L S M {C= ; ε= }. - Second, we search for good values of the RBF kernel parameter σ usng the predetermned parameters C and ε, and evaluate 45 dfferent alternatves for σ={0.00; 0.0; 0.03; 0.05; 0.08; 0.; 0.3; 0.5; 0.8; ;.3;.5;.8; 2; 2.3; 2.5; 2.8; 3; 3.3; 3.5; 3.8; 4; 4.3; 4.5; 4.8; 5; 5.3; 5.5; 5.8; 6; 7; 8; 9; 0; 5; 20; 25; 50; 80; 00; 200; 300; 400; 500; 000}. The value of σ whch generates the model wth the lowest mean absolute error (MAE) n the valdaton set s defned as the base parameter for the kernel functon. As result, we now have heurstc startng values for the three parameters of the SVR model, C, ε and σ. - Thrd, we defne a grd around base parameters C, ε and σ, and retan the best combnaton of parameters to be the fnal values used n the SVR model. In our experments, we tred fve dfferent values for each parameter C, ε, σ (factors

7 A study on the ablty of Support Vector Regresson and Neural Networks 7 0.5, 0.75,,.25 and.5 over the ntal values), thus creatng a grd of 25 possble parameter settngs. The parameter canddate of the grd s selected by usng the lowest MAE on the valdaton set as before. The scheme for Lnear SVR s very smlar, but wthout consderng parameter σ. Thus, second step for base parameter σ s not carred out, and the thrd step nvolves only 25 dfferent combnatons for C and ε. (for addtonal detals see [0]). For NN models, we used the backpropagaton algorthm to tran multple canddates of multlayer perceptron (MLP) networks. The network topology was obtaned usng a grd search of dfferent hdden nodes {0, 2,, 20} and actvaton functons {sgmod; tanh} wth fxed number of nput and output nodes, selectng the archtecture wth the lowest MAE on the valdaton set. The fnal model was ntalzed 20 tmes usng an (3-8-) archtecture comprsed of 3 nput nodes, 8 hdden nodes and a sngle output node for t+ predctons, applyng a sgmod transfer functon between the nput and hdden layers, and a lnear functon between hdden and output layers. As for SVR models, we selected the network wth the lowest valdaton mean absolute error (MAE) to calculate the test error results. 3.3 Expermental Results and Dscusson To evaluate our models we used the root mean squared error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE). Test set errors obtaned usng SVR and MLP models for each one of the fve seres analyzed n ths paper are shown n Table. As can be seen from Table, RBF SVR has the best performance (denoted n bold) on a level tme seres supermposed wth whte nose (E) and addtve seasonalty (S A ) patterns across all error measures. Lnear SVR s the best method for predctng lnear trend (T L ) and lnear trend wth multplcatve seasonalty patterns (T L S M ), whle MLPs provde best results for lnear trends wth addtve seasonalty (T L S A ) pattern. Table. Forecastng accuracy on the test set for RBF and lnear SVR models and MLP Seres RBF SVR Lnear SVR MLP RMSE MAE MAPE RMSE MAE MAPE RMSE MAE MAPE Seres E Seres S A Seres T L Seres T L S A Seres T L S M Sum Snce we evaluated artfcally constructed tme seres we can estmate the part of the forecastng errors caused by the artfcally created nose, whch due to ts random nature cannot be forecasted. Ths permts an analyss to what extent each method was capable of separatng nose from structure of varyng complexty on the unbased error measure of MAE. In applyng the true mean of the Gaussan resduals

8 8 Sven F. Crone, Jose Guajardo and Rchard Weber as an optmal forecast, we estmate a MAE of 3.80 as a lower bound forecast error for all tme seres on the test set. It becomes apparent, that RBF SVR exceeds even a perfect forecast for seres E and S A, whch can be attrbuted to the randomness of the data nherent n all ex ante evaluatons of forecastng experments. In contrast, RBF SVR sgnfcantly underperform on trended tme seres patterns, ndcatng nadequaces of the chosen kernel functon. On the contrary, lnear SVR shows a more robust predcton of all tme seres patterns. As the forecasts devates only slghtly from the lower bound n comparson to the level of the tme seres, as would be reflected n the MAPE, SVR wth lnear kernel functons may be consdered a robust method n forecastng arbtrary tme seres patterns wthout preprocessng. Smlarly, MLPs forecast all tme seres patterns robustly and wthout preprocessng wth a comparatve hgh accuracy close to lnear SVR and the lower bound. In summarzng over all tme seres, applyng an equal weght to each of the tme seres patterns, MLPs robustly outperform RBF SVR on all three error measures of MAE, MAPE and RMSE, whereas MLPs also moderately outperform lnear SVR. Ths ndcates that whle partcular kernel functons enable the SVR to outperform alternatve parameterzatons, MLPs or lnear SVR may prove a more robust alternatve n usng a sngle method to forecast a set of tme seres of dfferent patterns. In addton to these dstance based error measures, we evaluate the relatve performance by rankng each method by the ndvdual error measure, provded n Table 2. Table 2. Forecastng accuracy measured by ranks of methods for each error measure Rank by RMSE Rank by MAE Rank by MAPE SVR SVR MLP SVR SVR MLP SVR SVR MLP RBF lnear RBF lnear RBF lnear Seres E Seres S A Seres T L Seres T L S A Seres T L S M Sum of Ranks The fndngs by ranked error measures confrm lttle dfferences between lnear SVR and MLPs, wth MLPs provdng the best results for the lmted test desgn provded across all error measures. SVR wth RBF kernel, the most frequently used mplementaton n tme seres predcton wth SVR to date, performs sgnfcantly worse than the other methods. As must be expected, dfferent error measures dentfy dfferent best methods. In partcular, RMSE and MAPE are consdered to be based error measures. To lmt bases n the absence of a true objectve functon whch could motvate the use of a partcular error measure, we assume equal weght to each error and focus our conclusons on the MAE. To confrm the results of model accuracy from a statstcal pont of vew, we performed a pared-samples t test on the absolute values of the resduals over the test set data ponts. Results obtaned show that dfferences between

9 A study on the ablty of Support Vector Regresson and Neural Networks 9 model errors are statstcally sgnfcant when comparng RBF SVR to Lnear SVR (t=7.337; df=239; p<0.00), and RBF SVR to NN (t=6.999; df=239; p<0.00), although not when comparng NN to Lnear SVR (t=-0.989; df=239; p=0.324). Ths ndcates that no sgnfcant dfference n forecastng accuracy between the methods of lnear SVR and MLP may be derved from these experments. Consequently, we need to extend ths evaluaton on addtonal tme seres and varatons of MLPs. Results suggest that RBF SVR can predct seasonal patterns but no trends, whle lnear SVR and NN seem to be able to extrapolate trend as well as seasonal patterns accurately and wthout preprocessng. By examnng the resduals of the models, t can be observed that RBF SVR systematcally underestmate hold-out sample observatons for trended seres, whch corresponds to saturaton effects. 4 Concluson We have examned the ablty of RBF SVR, lnear SVR and MLP for predctng fve basc artfcal tme seres patterns: statonary, seasonalty, lnear trends, lnear trend wth addtve seasonalty, and lnear trend wth multplcatve seasonalty. Results obtaned usng multple error measures show that whle RBF SVR outperform other methods on non-trended data, they do not provde robust results across all patterns. For tme seres wth trend components, lnear SVR and MLP sgnfcantly outperform RBF SVR models, whch severely underestmate out-of-sample observatons, consstently laggng behnd upward trends. RBF SVR errors have shown to be statstcally sgnfcantly hgher than lnear SVR and NN errors. MLP demonstrate robust performance, provdng the hghest overall forecastng accuracy n across tme seres and dfferent statstcal error measures and rank based metrcs. Our results confrm prevous fndngs by Guajardo et. al [0], demonstratng accurate forecasts of seasonal tme seres wthout trends usng RBF SVR, even outperformng establshed statstcal methods such as ARIMAX. Also, they confrm results by Hansen et. al [9], who accurately predcted both lnear and nonlnear trends usng SVR, outperformng other methods such as ARIMA on several patterns. We assume that Hansen et al. also used lnear kernels, as they dd not fully document the kernel functons appled. A prelmnary hypothess for our poor results obtaned wth RBF SVR n extrapolatng trend patterns les n the lnear nature of ths trend. Prevous publcatons report smlar problems of closely related RBF-neural networks n predctng trends and nstatonary tme seres. Whle SVR wth lnear kernel functons and MLP wth lnear actvaton functons n the output unts may be partcularly suted to extrapolate lnear trends, we dd not conduct experments as to ther ablty to extrapolate non-lnear trends. These ssues wll be evaluated n an extended set of experments currently under nvestgaton by the authors, ncreasng the number of tme seres patterns and consderng addtonal knds of trend patterns, also evaluatng results aganst establshed statstcal forecastng methods as benchmarks. Addtonally, we wll

10 0 Sven F. Crone, Jose Guajardo and Rchard Weber evaluate the nfluence of preprocessng procedures such as deseasonalzaton to evaluate alternatve perspectves on the problem of extrapolatng tme seres patterns. Acknowledgement: Ths work has been supported n part by the Mllennum Nucleus Complex Engneerng Systems ( References. K. P. Lao and R. Fldes, The accuracy of a procedural approach to specfyng feedforward neural networks for forecastng, Computers & Operatons Research 32 (8) (2005) G.P. Zhang, B.E. Patuwo, and M.Y. Hu, Forecastng wth artfcal neural networks: The state of the art, Internatonal Journal of Forecastng,, 4, (998) 3. G.P. Zhang and M. Q, Neural network forecastng for seasonal and trend tme seres, European Journal of Operatonal Research 60, (2005). 4. L. J. Tashman, Out-of-sample tests of forecastng accuracy: an analyss and revew, Internatonal Journal of Forecastng 6 (4) (2000) V.Vapnk, The nature of statstcal learnng theory (Sprnger, New York, 995). 6. A.J. Smola and B. Schölkopf, A Tutoral on Support Vector Regresson, NeuroCOLT Techncal Report NC-TR , 998 (Royal Holloway College, Unversty of London, UK). 7. K. Müller, A. Smola, G. Rätsch, B. Schölkopf, J. Kohlmorgen, and V. Vapnk, n: Advances n Kernel Methods: Support Vector Learnng/ Usng Support Vector Machnes for Tme Seres Predcton, edted by B. Schölkopf, J. Burges, and A. Smola (MIT Press, 999), pp V.Vapnk, Statstcal Learnng Theory (John Wley and Sons, New York, 998). 9. J.V. Hansen, J.B. McDonald, and R.D. Nelson, Some evdence on forecastng tme-seres wth Support Vector Machnes, Journal of the Operatonal Research Socety,, -, J. Guajardo, J. Mranda, and R. Weber, A Hybrd Forecastng Methodology usng Feature Selecton and Support Vector Regresson, 5 th Internatonal Conference on Hybrd Intellgent Systems HIS 2005 (Ro de Janero, Brazl, 2005), pp C. M. Bshop, Neural networks for pattern recognton. Clarendon Press; Oxford Unversty Press, Oxford, S. Haykn, Neural networks: a comprehensve foundaton, 2nd ed. Prentce Hall, Upper Saddle Rver, N.J., A. Lapedes, R. Farber, and Los Alamos Natonal Laboratory, Nonlnear sgnal processng usng neural networks: predcton and system modellng, Los Alamos Natonal Laboratory, Los Alamos, N.M. LA-UR , V. Cherkassky and Y. Ma, Practcal selecton of SVM parameters and nose estmaton for SVM regresson, Neural Networks 7(), 3-26 (2004). 5. D. Mattera and S. Haykn, n: Advances n Kernel Methods: Support Vector Learnng/ Support Vector Machnes for Dynamc Reconstructon of a Chaotc System, edted by B. Schölkopf, J. Burges and A. Smola (MIT Press, 999), pp

Forecasting the Direction and Strength of Stock Market Movement

Forecasting the Direction and Strength of Stock Market Movement Forecastng the Drecton and Strength of Stock Market Movement Jngwe Chen Mng Chen Nan Ye cjngwe@stanford.edu mchen5@stanford.edu nanye@stanford.edu Abstract - Stock market s one of the most complcated systems

More information

An Evaluation of the Extended Logistic, Simple Logistic, and Gompertz Models for Forecasting Short Lifecycle Products and Services

An Evaluation of the Extended Logistic, Simple Logistic, and Gompertz Models for Forecasting Short Lifecycle Products and Services An Evaluaton of the Extended Logstc, Smple Logstc, and Gompertz Models for Forecastng Short Lfecycle Products and Servces Charles V. Trappey a,1, Hsn-yng Wu b a Professor (Management Scence), Natonal Chao

More information

The Development of Web Log Mining Based on Improve-K-Means Clustering Analysis

The Development of Web Log Mining Based on Improve-K-Means Clustering Analysis The Development of Web Log Mnng Based on Improve-K-Means Clusterng Analyss TngZhong Wang * College of Informaton Technology, Luoyang Normal Unversty, Luoyang, 471022, Chna wangtngzhong2@sna.cn Abstract.

More information

Can Auto Liability Insurance Purchases Signal Risk Attitude?

Can Auto Liability Insurance Purchases Signal Risk Attitude? Internatonal Journal of Busness and Economcs, 2011, Vol. 10, No. 2, 159-164 Can Auto Lablty Insurance Purchases Sgnal Rsk Atttude? Chu-Shu L Department of Internatonal Busness, Asa Unversty, Tawan Sheng-Chang

More information

What is Candidate Sampling

What is Candidate Sampling What s Canddate Samplng Say we have a multclass or mult label problem where each tranng example ( x, T ) conssts of a context x a small (mult)set of target classes T out of a large unverse L of possble

More information

Modelling of Web Domain Visits by Radial Basis Function Neural Networks and Support Vector Machine Regression

Modelling of Web Domain Visits by Radial Basis Function Neural Networks and Support Vector Machine Regression Modellng of Web Doman Vsts by Radal Bass Functon Neural Networks and Support Vector Machne Regresson Vladmír Olej, Jana Flpová Insttute of System Engneerng and Informatcs Faculty of Economcs and Admnstraton,

More information

Forecasting the Demand of Emergency Supplies: Based on the CBR Theory and BP Neural Network

Forecasting the Demand of Emergency Supplies: Based on the CBR Theory and BP Neural Network 700 Proceedngs of the 8th Internatonal Conference on Innovaton & Management Forecastng the Demand of Emergency Supples: Based on the CBR Theory and BP Neural Network Fu Deqang, Lu Yun, L Changbng School

More information

benefit is 2, paid if the policyholder dies within the year, and probability of death within the year is ).

benefit is 2, paid if the policyholder dies within the year, and probability of death within the year is ). REVIEW OF RISK MANAGEMENT CONCEPTS LOSS DISTRIBUTIONS AND INSURANCE Loss and nsurance: When someone s subject to the rsk of ncurrng a fnancal loss, the loss s generally modeled usng a random varable or

More information

Statistical Methods to Develop Rating Models

Statistical Methods to Develop Rating Models Statstcal Methods to Develop Ratng Models [Evelyn Hayden and Danel Porath, Österrechsche Natonalbank and Unversty of Appled Scences at Manz] Source: The Basel II Rsk Parameters Estmaton, Valdaton, and

More information

Single and multiple stage classifiers implementing logistic discrimination

Single and multiple stage classifiers implementing logistic discrimination Sngle and multple stage classfers mplementng logstc dscrmnaton Hélo Radke Bttencourt 1 Dens Alter de Olvera Moraes 2 Vctor Haertel 2 1 Pontfíca Unversdade Católca do Ro Grande do Sul - PUCRS Av. Ipranga,

More information

How To Understand The Results Of The German Meris Cloud And Water Vapour Product

How To Understand The Results Of The German Meris Cloud And Water Vapour Product Ttel: Project: Doc. No.: MERIS level 3 cloud and water vapour products MAPP MAPP-ATBD-ClWVL3 Issue: 1 Revson: 0 Date: 9.12.1998 Functon Name Organsaton Sgnature Date Author: Bennartz FUB Preusker FUB Schüller

More information

Bayesian Network Based Causal Relationship Identification and Funding Success Prediction in P2P Lending

Bayesian Network Based Causal Relationship Identification and Funding Success Prediction in P2P Lending Proceedngs of 2012 4th Internatonal Conference on Machne Learnng and Computng IPCSIT vol. 25 (2012) (2012) IACSIT Press, Sngapore Bayesan Network Based Causal Relatonshp Identfcaton and Fundng Success

More information

CS 2750 Machine Learning. Lecture 3. Density estimation. CS 2750 Machine Learning. Announcements

CS 2750 Machine Learning. Lecture 3. Density estimation. CS 2750 Machine Learning. Announcements Lecture 3 Densty estmaton Mlos Hauskrecht mlos@cs.ptt.edu 5329 Sennott Square Next lecture: Matlab tutoral Announcements Rules for attendng the class: Regstered for credt Regstered for audt (only f there

More information

Multiple-Period Attribution: Residuals and Compounding

Multiple-Period Attribution: Residuals and Compounding Multple-Perod Attrbuton: Resduals and Compoundng Our revewer gave these authors full marks for dealng wth an ssue that performance measurers and vendors often regard as propretary nformaton. In 1994, Dens

More information

Development of an intelligent system for tool wear monitoring applying neural networks

Development of an intelligent system for tool wear monitoring applying neural networks of Achevements n Materals and Manufacturng Engneerng VOLUME 14 ISSUE 1-2 January-February 2006 Development of an ntellgent system for tool wear montorng applyng neural networks A. Antć a, J. Hodolč a,

More information

An Efficient and Simplified Model for Forecasting using SRM

An Efficient and Simplified Model for Forecasting using SRM HAFIZ MUHAMMAD SHAHZAD ASIF*, MUHAMMAD FAISAL HAYAT*, AND TAUQIR AHMAD* RECEIVED ON 15.04.013 ACCEPTED ON 09.01.014 ABSTRACT Learnng form contnuous fnancal systems play a vtal role n enterprse operatons.

More information

Institute of Informatics, Faculty of Business and Management, Brno University of Technology,Czech Republic

Institute of Informatics, Faculty of Business and Management, Brno University of Technology,Czech Republic Lagrange Multplers as Quanttatve Indcators n Economcs Ivan Mezník Insttute of Informatcs, Faculty of Busness and Management, Brno Unversty of TechnologCzech Republc Abstract The quanttatve role of Lagrange

More information

IMPACT ANALYSIS OF A CELLULAR PHONE

IMPACT ANALYSIS OF A CELLULAR PHONE 4 th ASA & μeta Internatonal Conference IMPACT AALYSIS OF A CELLULAR PHOE We Lu, 2 Hongy L Bejng FEAonlne Engneerng Co.,Ltd. Bejng, Chna ABSTRACT Drop test smulaton plays an mportant role n nvestgatng

More information

An Interest-Oriented Network Evolution Mechanism for Online Communities

An Interest-Oriented Network Evolution Mechanism for Online Communities An Interest-Orented Network Evoluton Mechansm for Onlne Communtes Cahong Sun and Xaopng Yang School of Informaton, Renmn Unversty of Chna, Bejng 100872, P.R. Chna {chsun,yang}@ruc.edu.cn Abstract. Onlne

More information

NEURO-FUZZY INFERENCE SYSTEM FOR E-COMMERCE WEBSITE EVALUATION

NEURO-FUZZY INFERENCE SYSTEM FOR E-COMMERCE WEBSITE EVALUATION NEURO-FUZZY INFERENE SYSTEM FOR E-OMMERE WEBSITE EVALUATION Huan Lu, School of Software, Harbn Unversty of Scence and Technology, Harbn, hna Faculty of Appled Mathematcs and omputer Scence, Belarusan State

More information

An Alternative Way to Measure Private Equity Performance

An Alternative Way to Measure Private Equity Performance An Alternatve Way to Measure Prvate Equty Performance Peter Todd Parlux Investment Technology LLC Summary Internal Rate of Return (IRR) s probably the most common way to measure the performance of prvate

More information

Abstract. 260 Business Intelligence Journal July IDENTIFICATION OF DEMAND THROUGH STATISTICAL DISTRIBUTION MODELING FOR IMPROVED DEMAND FORECASTING

Abstract. 260 Business Intelligence Journal July IDENTIFICATION OF DEMAND THROUGH STATISTICAL DISTRIBUTION MODELING FOR IMPROVED DEMAND FORECASTING 260 Busness Intellgence Journal July IDENTIFICATION OF DEMAND THROUGH STATISTICAL DISTRIBUTION MODELING FOR IMPROVED DEMAND FORECASTING Murphy Choy Mchelle L.F. Cheong School of Informaton Systems, Sngapore

More information

Time Delayed Independent Component Analysis for Data Quality Monitoring

Time Delayed Independent Component Analysis for Data Quality Monitoring IWSSIP 1-17th Internatonal Conference on Systems, Sgnals and Image Processng Tme Delayed Independent Component Analyss for Data Qualty Montorng José Márco Faer Sgnal Processng Laboratory, COE/Pol Federal

More information

Exhaustive Regression. An Exploration of Regression-Based Data Mining Techniques Using Super Computation

Exhaustive Regression. An Exploration of Regression-Based Data Mining Techniques Using Super Computation Exhaustve Regresson An Exploraton of Regresson-Based Data Mnng Technques Usng Super Computaton Antony Daves, Ph.D. Assocate Professor of Economcs Duquesne Unversty Pttsburgh, PA 58 Research Fellow The

More information

L10: Linear discriminants analysis

L10: Linear discriminants analysis L0: Lnear dscrmnants analyss Lnear dscrmnant analyss, two classes Lnear dscrmnant analyss, C classes LDA vs. PCA Lmtatons of LDA Varants of LDA Other dmensonalty reducton methods CSCE 666 Pattern Analyss

More information

ANALYZING THE RELATIONSHIPS BETWEEN QUALITY, TIME, AND COST IN PROJECT MANAGEMENT DECISION MAKING

ANALYZING THE RELATIONSHIPS BETWEEN QUALITY, TIME, AND COST IN PROJECT MANAGEMENT DECISION MAKING ANALYZING THE RELATIONSHIPS BETWEEN QUALITY, TIME, AND COST IN PROJECT MANAGEMENT DECISION MAKING Matthew J. Lberatore, Department of Management and Operatons, Vllanova Unversty, Vllanova, PA 19085, 610-519-4390,

More information

Support Vector Machines

Support Vector Machines Support Vector Machnes Max Wellng Department of Computer Scence Unversty of Toronto 10 Kng s College Road Toronto, M5S 3G5 Canada wellng@cs.toronto.edu Abstract Ths s a note to explan support vector machnes.

More information

Feature selection for intrusion detection. Slobodan Petrović NISlab, Gjøvik University College

Feature selection for intrusion detection. Slobodan Petrović NISlab, Gjøvik University College Feature selecton for ntruson detecton Slobodan Petrovć NISlab, Gjøvk Unversty College Contents The feature selecton problem Intruson detecton Traffc features relevant for IDS The CFS measure The mrmr measure

More information

How To Calculate The Accountng Perod Of Nequalty

How To Calculate The Accountng Perod Of Nequalty Inequalty and The Accountng Perod Quentn Wodon and Shlomo Ytzha World Ban and Hebrew Unversty September Abstract Income nequalty typcally declnes wth the length of tme taen nto account for measurement.

More information

Logistic Regression. Lecture 4: More classifiers and classes. Logistic regression. Adaboost. Optimization. Multiple class classification

Logistic Regression. Lecture 4: More classifiers and classes. Logistic regression. Adaboost. Optimization. Multiple class classification Lecture 4: More classfers and classes C4B Machne Learnng Hlary 20 A. Zsserman Logstc regresson Loss functons revsted Adaboost Loss functons revsted Optmzaton Multple class classfcaton Logstc Regresson

More information

Causal, Explanatory Forecasting. Analysis. Regression Analysis. Simple Linear Regression. Which is Independent? Forecasting

Causal, Explanatory Forecasting. Analysis. Regression Analysis. Simple Linear Regression. Which is Independent? Forecasting Causal, Explanatory Forecastng Assumes cause-and-effect relatonshp between system nputs and ts output Forecastng wth Regresson Analyss Rchard S. Barr Inputs System Cause + Effect Relatonshp The job of

More information

Risk-based Fatigue Estimate of Deep Water Risers -- Course Project for EM388F: Fracture Mechanics, Spring 2008

Risk-based Fatigue Estimate of Deep Water Risers -- Course Project for EM388F: Fracture Mechanics, Spring 2008 Rsk-based Fatgue Estmate of Deep Water Rsers -- Course Project for EM388F: Fracture Mechancs, Sprng 2008 Chen Sh Department of Cvl, Archtectural, and Envronmental Engneerng The Unversty of Texas at Austn

More information

Gender Classification for Real-Time Audience Analysis System

Gender Classification for Real-Time Audience Analysis System Gender Classfcaton for Real-Tme Audence Analyss System Vladmr Khryashchev, Lev Shmaglt, Andrey Shemyakov, Anton Lebedev Yaroslavl State Unversty Yaroslavl, Russa vhr@yandex.ru, shmaglt_lev@yahoo.com, andrey.shemakov@gmal.com,

More information

Forecasting and Modelling Electricity Demand Using Anfis Predictor

Forecasting and Modelling Electricity Demand Using Anfis Predictor Journal of Mathematcs and Statstcs 7 (4): 75-8, 0 ISSN 549-3644 0 Scence Publcatons Forecastng and Modellng Electrcty Demand Usng Anfs Predctor M. Mordjaou and B. Boudjema Department of Electrcal Engneerng,

More information

Module 2 LOSSLESS IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur

Module 2 LOSSLESS IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur Module LOSSLESS IMAGE COMPRESSION SYSTEMS Lesson 3 Lossless Compresson: Huffman Codng Instructonal Objectves At the end of ths lesson, the students should be able to:. Defne and measure source entropy..

More information

Calendar Corrected Chaotic Forecast of Financial Time Series

Calendar Corrected Chaotic Forecast of Financial Time Series INTERNATIONAL JOURNAL OF BUSINESS, 11(4), 2006 ISSN: 1083 4346 Calendar Corrected Chaotc Forecast of Fnancal Tme Seres Alexandros Leonttss a and Costas Sropoulos b a Center for Research and Applcatons

More information

THE APPLICATION OF DATA MINING TECHNIQUES AND MULTIPLE CLASSIFIERS TO MARKETING DECISION

THE APPLICATION OF DATA MINING TECHNIQUES AND MULTIPLE CLASSIFIERS TO MARKETING DECISION Internatonal Journal of Electronc Busness Management, Vol. 3, No. 4, pp. 30-30 (2005) 30 THE APPLICATION OF DATA MINING TECHNIQUES AND MULTIPLE CLASSIFIERS TO MARKETING DECISION Yu-Mn Chang *, Yu-Cheh

More information

Staff Paper. Farm Savings Accounts: Examining Income Variability, Eligibility, and Benefits. Brent Gloy, Eddy LaDue, and Charles Cuykendall

Staff Paper. Farm Savings Accounts: Examining Income Variability, Eligibility, and Benefits. Brent Gloy, Eddy LaDue, and Charles Cuykendall SP 2005-02 August 2005 Staff Paper Department of Appled Economcs and Management Cornell Unversty, Ithaca, New York 14853-7801 USA Farm Savngs Accounts: Examnng Income Varablty, Elgblty, and Benefts Brent

More information

Predicting Software Development Project Outcomes *

Predicting Software Development Project Outcomes * Predctng Software Development Project Outcomes * Rosna Weber, Mchael Waller, June Verner, Wllam Evanco College of Informaton Scence & Technology, Drexel Unversty 3141 Chestnut Street Phladelpha, PA 19104

More information

Calculation of Sampling Weights

Calculation of Sampling Weights Perre Foy Statstcs Canada 4 Calculaton of Samplng Weghts 4.1 OVERVIEW The basc sample desgn used n TIMSS Populatons 1 and 2 was a two-stage stratfed cluster desgn. 1 The frst stage conssted of a sample

More information

The Current Employment Statistics (CES) survey,

The Current Employment Statistics (CES) survey, Busness Brths and Deaths Impact of busness brths and deaths n the payroll survey The CES probablty-based sample redesgn accounts for most busness brth employment through the mputaton of busness deaths,

More information

CHOLESTEROL REFERENCE METHOD LABORATORY NETWORK. Sample Stability Protocol

CHOLESTEROL REFERENCE METHOD LABORATORY NETWORK. Sample Stability Protocol CHOLESTEROL REFERENCE METHOD LABORATORY NETWORK Sample Stablty Protocol Background The Cholesterol Reference Method Laboratory Network (CRMLN) developed certfcaton protocols for total cholesterol, HDL

More information

PSYCHOLOGICAL RESEARCH (PYC 304-C) Lecture 12

PSYCHOLOGICAL RESEARCH (PYC 304-C) Lecture 12 14 The Ch-squared dstrbuton PSYCHOLOGICAL RESEARCH (PYC 304-C) Lecture 1 If a normal varable X, havng mean µ and varance σ, s standardsed, the new varable Z has a mean 0 and varance 1. When ths standardsed

More information

Analysis of Premium Liabilities for Australian Lines of Business

Analysis of Premium Liabilities for Australian Lines of Business Summary of Analyss of Premum Labltes for Australan Lnes of Busness Emly Tao Honours Research Paper, The Unversty of Melbourne Emly Tao Acknowledgements I am grateful to the Australan Prudental Regulaton

More information

The Application of Fractional Brownian Motion in Option Pricing

The Application of Fractional Brownian Motion in Option Pricing Vol. 0, No. (05), pp. 73-8 http://dx.do.org/0.457/jmue.05.0..6 The Applcaton of Fractonal Brownan Moton n Opton Prcng Qng-xn Zhou School of Basc Scence,arbn Unversty of Commerce,arbn zhouqngxn98@6.com

More information

DEFINING %COMPLETE IN MICROSOFT PROJECT

DEFINING %COMPLETE IN MICROSOFT PROJECT CelersSystems DEFINING %COMPLETE IN MICROSOFT PROJECT PREPARED BY James E Aksel, PMP, PMI-SP, MVP For Addtonal Informaton about Earned Value Management Systems and reportng, please contact: CelersSystems,

More information

Project Networks With Mixed-Time Constraints

Project Networks With Mixed-Time Constraints Project Networs Wth Mxed-Tme Constrants L Caccetta and B Wattananon Western Australan Centre of Excellence n Industral Optmsaton (WACEIO) Curtn Unversty of Technology GPO Box U1987 Perth Western Australa

More information

APPLICATION OF PROBE DATA COLLECTED VIA INFRARED BEACONS TO TRAFFIC MANEGEMENT

APPLICATION OF PROBE DATA COLLECTED VIA INFRARED BEACONS TO TRAFFIC MANEGEMENT APPLICATION OF PROBE DATA COLLECTED VIA INFRARED BEACONS TO TRAFFIC MANEGEMENT Toshhko Oda (1), Kochro Iwaoka (2) (1), (2) Infrastructure Systems Busness Unt, Panasonc System Networks Co., Ltd. Saedo-cho

More information

DAILY CRUDE OIL PRICE FORECASTING MODEL USING ARIMA, GENERALIZED AUTOREGRESSIVE CONDITIONAL HETEROSCEDASTIC AND SUPPORT VECTOR MACHINES

DAILY CRUDE OIL PRICE FORECASTING MODEL USING ARIMA, GENERALIZED AUTOREGRESSIVE CONDITIONAL HETEROSCEDASTIC AND SUPPORT VECTOR MACHINES Amercan Journal of Appled Scences (3): 45-43, 04 ISS: 546-939 04 Scence Publcaton do:0.3844/ajassp.04.45.43 Publshed Onlne (3) 04 (http://www.thescpub.com/ajas.toc) DAILY CRUDE OIL PRICE FORECASTIG MODEL

More information

Study on Model of Risks Assessment of Standard Operation in Rural Power Network

Study on Model of Risks Assessment of Standard Operation in Rural Power Network Study on Model of Rsks Assessment of Standard Operaton n Rural Power Network Qngj L 1, Tao Yang 2 1 Qngj L, College of Informaton and Electrcal Engneerng, Shenyang Agrculture Unversty, Shenyang 110866,

More information

Course outline. Financial Time Series Analysis. Overview. Data analysis. Predictive signal. Trading strategy

Course outline. Financial Time Series Analysis. Overview. Data analysis. Predictive signal. Trading strategy Fnancal Tme Seres Analyss Patrck McSharry patrck@mcsharry.net www.mcsharry.net Trnty Term 2014 Mathematcal Insttute Unversty of Oxford Course outlne 1. Data analyss, probablty, correlatons, vsualsaton

More information

How To Know The Components Of Mean Squared Error Of Herarchcal Estmator S

How To Know The Components Of Mean Squared Error Of Herarchcal Estmator S S C H E D A E I N F O R M A T I C A E VOLUME 0 0 On Mean Squared Error of Herarchcal Estmator Stans law Brodowsk Faculty of Physcs, Astronomy, and Appled Computer Scence, Jagellonan Unversty, Reymonta

More information

Design and Development of a Security Evaluation Platform Based on International Standards

Design and Development of a Security Evaluation Platform Based on International Standards Internatonal Journal of Informatcs Socety, VOL.5, NO.2 (203) 7-80 7 Desgn and Development of a Securty Evaluaton Platform Based on Internatonal Standards Yuj Takahash and Yoshm Teshgawara Graduate School

More information

Statistical algorithms in Review Manager 5

Statistical algorithms in Review Manager 5 Statstcal algorthms n Reve Manager 5 Jonathan J Deeks and Julan PT Hggns on behalf of the Statstcal Methods Group of The Cochrane Collaboraton August 00 Data structure Consder a meta-analyss of k studes

More information

Intra-day Trading of the FTSE-100 Futures Contract Using Neural Networks With Wavelet Encodings

Intra-day Trading of the FTSE-100 Futures Contract Using Neural Networks With Wavelet Encodings Submtted to European Journal of Fnance Intra-day Tradng of the FTSE-00 Futures Contract Usng eural etworks Wth Wavelet Encodngs D L Toulson S P Toulson Intellgent Fnancal Systems Lmted Sute 4 Greener House

More information

Damage detection in composite laminates using coin-tap method

Damage detection in composite laminates using coin-tap method Damage detecton n composte lamnates usng con-tap method S.J. Km Korea Aerospace Research Insttute, 45 Eoeun-Dong, Youseong-Gu, 35-333 Daejeon, Republc of Korea yaeln@kar.re.kr 45 The con-tap test has the

More information

Lecture 2: Single Layer Perceptrons Kevin Swingler

Lecture 2: Single Layer Perceptrons Kevin Swingler Lecture 2: Sngle Layer Perceptrons Kevn Sngler kms@cs.str.ac.uk Recap: McCulloch-Ptts Neuron Ths vastly smplfed model of real neurons s also knon as a Threshold Logc Unt: W 2 A Y 3 n W n. A set of synapses

More information

Joint Scheduling of Processing and Shuffle Phases in MapReduce Systems

Joint Scheduling of Processing and Shuffle Phases in MapReduce Systems Jont Schedulng of Processng and Shuffle Phases n MapReduce Systems Fangfe Chen, Mural Kodalam, T. V. Lakshman Department of Computer Scence and Engneerng, The Penn State Unversty Bell Laboratores, Alcatel-Lucent

More information

HOUSEHOLDS DEBT BURDEN: AN ANALYSIS BASED ON MICROECONOMIC DATA*

HOUSEHOLDS DEBT BURDEN: AN ANALYSIS BASED ON MICROECONOMIC DATA* HOUSEHOLDS DEBT BURDEN: AN ANALYSIS BASED ON MICROECONOMIC DATA* Luísa Farnha** 1. INTRODUCTION The rapd growth n Portuguese households ndebtedness n the past few years ncreased the concerns that debt

More information

ECE544NA Final Project: Robust Machine Learning Hardware via Classifier Ensemble

ECE544NA Final Project: Robust Machine Learning Hardware via Classifier Ensemble 1 ECE544NA Fnal Project: Robust Machne Learnng Hardware va Classfer Ensemble Sa Zhang, szhang12@llnos.edu Dept. of Electr. & Comput. Eng., Unv. of Illnos at Urbana-Champagn, Urbana, IL, USA Abstract In

More information

The Journal of Systems and Software

The Journal of Systems and Software The Journal of Systems and Software 82 (2009) 241 252 Contents lsts avalable at ScenceDrect The Journal of Systems and Software journal homepage: www. elsever. com/ locate/ jss A study of project selecton

More information

On the Use of Neural Network as a Universal Approximator

On the Use of Neural Network as a Universal Approximator Internatonal Journal of Scences and Technques of Automatc control & computer engneerng IJ-STA, Volume, N, Jul 8, pp 386 399 On the Use of Neural Network as a Unversal Appromator Amel SIFAOUI, Afef ABDELKRIM,

More information

On the Optimal Control of a Cascade of Hydro-Electric Power Stations

On the Optimal Control of a Cascade of Hydro-Electric Power Stations On the Optmal Control of a Cascade of Hydro-Electrc Power Statons M.C.M. Guedes a, A.F. Rbero a, G.V. Smrnov b and S. Vlela c a Department of Mathematcs, School of Scences, Unversty of Porto, Portugal;

More information

A DATA MINING APPLICATION IN A STUDENT DATABASE

A DATA MINING APPLICATION IN A STUDENT DATABASE JOURNAL OF AERONAUTICS AND SPACE TECHNOLOGIES JULY 005 VOLUME NUMBER (53-57) A DATA MINING APPLICATION IN A STUDENT DATABASE Şenol Zafer ERDOĞAN Maltepe Ünversty Faculty of Engneerng Büyükbakkalköy-Istanbul

More information

MATHEMATICAL ENGINEERING TECHNICAL REPORTS. Sequential Optimizing Investing Strategy with Neural Networks

MATHEMATICAL ENGINEERING TECHNICAL REPORTS. Sequential Optimizing Investing Strategy with Neural Networks MATHEMATICAL ENGINEERING TECHNICAL REPORTS Sequental Optmzng Investng Strategy wth Neural Networks Ryo ADACHI and Akmch TAKEMURA METR 2010 03 February 2010 DEPARTMENT OF MATHEMATICAL INFORMATICS GRADUATE

More information

A DYNAMIC CRASHING METHOD FOR PROJECT MANAGEMENT USING SIMULATION-BASED OPTIMIZATION. Michael E. Kuhl Radhamés A. Tolentino-Peña

A DYNAMIC CRASHING METHOD FOR PROJECT MANAGEMENT USING SIMULATION-BASED OPTIMIZATION. Michael E. Kuhl Radhamés A. Tolentino-Peña Proceedngs of the 2008 Wnter Smulaton Conference S. J. Mason, R. R. Hll, L. Mönch, O. Rose, T. Jefferson, J. W. Fowler eds. A DYNAMIC CRASHING METHOD FOR PROJECT MANAGEMENT USING SIMULATION-BASED OPTIMIZATION

More information

CHAPTER 14 MORE ABOUT REGRESSION

CHAPTER 14 MORE ABOUT REGRESSION CHAPTER 14 MORE ABOUT REGRESSION We learned n Chapter 5 that often a straght lne descrbes the pattern of a relatonshp between two quanttatve varables. For nstance, n Example 5.1 we explored the relatonshp

More information

How Sets of Coherent Probabilities May Serve as Models for Degrees of Incoherence

How Sets of Coherent Probabilities May Serve as Models for Degrees of Incoherence 1 st Internatonal Symposum on Imprecse Probabltes and Ther Applcatons, Ghent, Belgum, 29 June 2 July 1999 How Sets of Coherent Probabltes May Serve as Models for Degrees of Incoherence Mar J. Schervsh

More information

Power-of-Two Policies for Single- Warehouse Multi-Retailer Inventory Systems with Order Frequency Discounts

Power-of-Two Policies for Single- Warehouse Multi-Retailer Inventory Systems with Order Frequency Discounts Power-of-wo Polces for Sngle- Warehouse Mult-Retaler Inventory Systems wth Order Frequency Dscounts José A. Ventura Pennsylvana State Unversty (USA) Yale. Herer echnon Israel Insttute of echnology (Israel)

More information

Scale Dependence of Overconfidence in Stock Market Volatility Forecasts

Scale Dependence of Overconfidence in Stock Market Volatility Forecasts Scale Dependence of Overconfdence n Stoc Maret Volatlty Forecasts Marus Glaser, Thomas Langer, Jens Reynders, Martn Weber* June 7, 007 Abstract In ths study, we analyze whether volatlty forecasts (judgmental

More information

v a 1 b 1 i, a 2 b 2 i,..., a n b n i.

v a 1 b 1 i, a 2 b 2 i,..., a n b n i. SECTION 8.4 COMPLEX VECTOR SPACES AND INNER PRODUCTS 455 8.4 COMPLEX VECTOR SPACES AND INNER PRODUCTS All the vector spaces we have studed thus far n the text are real vector spaces snce the scalars are

More information

León, Monterrey, Nuevo León, MÉXICO 4

León, Monterrey, Nuevo León, MÉXICO 4 Statstcal Characterzaton and Optmzaton of Artfcal Neural Networks n Tme Seres Forecastng: The One-Perod Forecast Case Caracterzacón Estadístca y Optmzacón de Redes Neuronales Artfcales para Pronóstco de

More information

Dynamic Pricing for Smart Grid with Reinforcement Learning

Dynamic Pricing for Smart Grid with Reinforcement Learning Dynamc Prcng for Smart Grd wth Renforcement Learnng Byung-Gook Km, Yu Zhang, Mhaela van der Schaar, and Jang-Won Lee Samsung Electroncs, Suwon, Korea Department of Electrcal Engneerng, UCLA, Los Angeles,

More information

Financial market forecasting using a two-step kernel learning method for the support vector regression

Financial market forecasting using a two-step kernel learning method for the support vector regression Ann Oper Res (2010) 174: 103 120 DOI 10.1007/s10479-008-0357-7 Fnancal market forecastng usng a two-step kernel learnng method for the support vector regresson L Wang J Zhu Publshed onlne: 28 May 2008

More information

A Novel Methodology of Working Capital Management for Large. Public Constructions by Using Fuzzy S-curve Regression

A Novel Methodology of Working Capital Management for Large. Public Constructions by Using Fuzzy S-curve Regression Novel Methodology of Workng Captal Management for Large Publc Constructons by Usng Fuzzy S-curve Regresson Cheng-Wu Chen, Morrs H. L. Wang and Tng-Ya Hseh Department of Cvl Engneerng, Natonal Central Unversty,

More information

Dynamic Resource Allocation for MapReduce with Partitioning Skew

Dynamic Resource Allocation for MapReduce with Partitioning Skew Ths artcle has been accepted for publcaton n a future ssue of ths journal, but has not been fully edted. Content may change pror to fnal publcaton. Ctaton nformaton: DOI 1.119/TC.216.253286, IEEE Transactons

More information

Performance Analysis and Coding Strategy of ECOC SVMs

Performance Analysis and Coding Strategy of ECOC SVMs Internatonal Journal of Grd and Dstrbuted Computng Vol.7, No. (04), pp.67-76 http://dx.do.org/0.457/jgdc.04.7..07 Performance Analyss and Codng Strategy of ECOC SVMs Zhgang Yan, and Yuanxuan Yang, School

More information

Different Methods of Long-Term Electric Load Demand Forecasting; A Comprehensive Review

Different Methods of Long-Term Electric Load Demand Forecasting; A Comprehensive Review Dfferent Methods of Long-Term Electrc Load Demand Forecastng; A Comprehensve Revew L. Ghods* and M. Kalantar* Abstract: Long-term demand forecastng presents the frst step n plannng and developng future

More information

AN APPOINTMENT ORDER OUTPATIENT SCHEDULING SYSTEM THAT IMPROVES OUTPATIENT EXPERIENCE

AN APPOINTMENT ORDER OUTPATIENT SCHEDULING SYSTEM THAT IMPROVES OUTPATIENT EXPERIENCE AN APPOINTMENT ORDER OUTPATIENT SCHEDULING SYSTEM THAT IMPROVES OUTPATIENT EXPERIENCE Yu-L Huang Industral Engneerng Department New Mexco State Unversty Las Cruces, New Mexco 88003, U.S.A. Abstract Patent

More information

8 Algorithm for Binary Searching in Trees

8 Algorithm for Binary Searching in Trees 8 Algorthm for Bnary Searchng n Trees In ths secton we present our algorthm for bnary searchng n trees. A crucal observaton employed by the algorthm s that ths problem can be effcently solved when the

More information

GRAVITY DATA VALIDATION AND OUTLIER DETECTION USING L 1 -NORM

GRAVITY DATA VALIDATION AND OUTLIER DETECTION USING L 1 -NORM GRAVITY DATA VALIDATION AND OUTLIER DETECTION USING L 1 -NORM BARRIOT Jean-Perre, SARRAILH Mchel BGI/CNES 18.av.E.Beln 31401 TOULOUSE Cedex 4 (France) Emal: jean-perre.barrot@cnes.fr 1/Introducton The

More information

A Hierarchical Anomaly Network Intrusion Detection System using Neural Network Classification

A Hierarchical Anomaly Network Intrusion Detection System using Neural Network Classification IDC IDC A Herarchcal Anomaly Network Intruson Detecton System usng Neural Network Classfcaton ZHENG ZHANG, JUN LI, C. N. MANIKOPOULOS, JAY JORGENSON and JOSE UCLES ECE Department, New Jersey Inst. of Tech.,

More information

Transition Matrix Models of Consumer Credit Ratings

Transition Matrix Models of Consumer Credit Ratings Transton Matrx Models of Consumer Credt Ratngs Abstract Although the corporate credt rsk lterature has many studes modellng the change n the credt rsk of corporate bonds over tme, there s far less analyss

More information

A Hybrid Model for Forecasting Sales in Turkish Paint Industry

A Hybrid Model for Forecasting Sales in Turkish Paint Industry Internatonal Journal of Computatonal Intellgence Systems, Vol.2, No. 3 (October, 2009), 277-287 A Hybrd Model for Forecastng Sales n Turksh Pant Industry Alp Ustundag * Department of Industral Engneerng,

More information

Characterization of Assembly. Variation Analysis Methods. A Thesis. Presented to the. Department of Mechanical Engineering. Brigham Young University

Characterization of Assembly. Variation Analysis Methods. A Thesis. Presented to the. Department of Mechanical Engineering. Brigham Young University Characterzaton of Assembly Varaton Analyss Methods A Thess Presented to the Department of Mechancal Engneerng Brgham Young Unversty In Partal Fulfllment of the Requrements for the Degree Master of Scence

More information

An Enhanced Super-Resolution System with Improved Image Registration, Automatic Image Selection, and Image Enhancement

An Enhanced Super-Resolution System with Improved Image Registration, Automatic Image Selection, and Image Enhancement An Enhanced Super-Resoluton System wth Improved Image Regstraton, Automatc Image Selecton, and Image Enhancement Yu-Chuan Kuo ( ), Chen-Yu Chen ( ), and Chou-Shann Fuh ( ) Department of Computer Scence

More information

Data Broadcast on a Multi-System Heterogeneous Overlayed Wireless Network *

Data Broadcast on a Multi-System Heterogeneous Overlayed Wireless Network * JOURNAL OF INFORMATION SCIENCE AND ENGINEERING 24, 819-840 (2008) Data Broadcast on a Mult-System Heterogeneous Overlayed Wreless Network * Department of Computer Scence Natonal Chao Tung Unversty Hsnchu,

More information

Properties of Indoor Received Signal Strength for WLAN Location Fingerprinting

Properties of Indoor Received Signal Strength for WLAN Location Fingerprinting Propertes of Indoor Receved Sgnal Strength for WLAN Locaton Fngerprntng Kamol Kaemarungs and Prashant Krshnamurthy Telecommuncatons Program, School of Informaton Scences, Unversty of Pttsburgh E-mal: kakst2,prashk@ptt.edu

More information

An MILP model for planning of batch plants operating in a campaign-mode

An MILP model for planning of batch plants operating in a campaign-mode An MILP model for plannng of batch plants operatng n a campagn-mode Yanna Fumero Insttuto de Desarrollo y Dseño CONICET UTN yfumero@santafe-concet.gov.ar Gabrela Corsano Insttuto de Desarrollo y Dseño

More information

New Approaches to Support Vector Ordinal Regression

New Approaches to Support Vector Ordinal Regression New Approaches to Support Vector Ordnal Regresson We Chu chuwe@gatsby.ucl.ac.uk Gatsby Computatonal Neuroscence Unt, Unversty College London, London, WCN 3AR, UK S. Sathya Keerth selvarak@yahoo-nc.com

More information

A COLLABORATIVE TRADING MODEL BY SUPPORT VECTOR REGRESSION AND TS FUZZY RULE FOR DAILY STOCK TURNING POINTS DETECTION

A COLLABORATIVE TRADING MODEL BY SUPPORT VECTOR REGRESSION AND TS FUZZY RULE FOR DAILY STOCK TURNING POINTS DETECTION A COLLABORATIVE TRADING MODEL BY SUPPORT VECTOR REGRESSION AND TS FUZZY RULE FOR DAILY STOCK TURNING POINTS DETECTION JHENG-LONG WU, PEI-CHANN CHANG, KAI-TING CHANG Department of Informaton Management,

More information

Marginal Benefit Incidence Analysis Using a Single Cross-section of Data. Mohamed Ihsan Ajwad and Quentin Wodon 1. World Bank.

Marginal Benefit Incidence Analysis Using a Single Cross-section of Data. Mohamed Ihsan Ajwad and Quentin Wodon 1. World Bank. Margnal Beneft Incdence Analyss Usng a Sngle Cross-secton of Data Mohamed Ihsan Ajwad and uentn Wodon World Bank August 200 Abstract In a recent paper, Lanjouw and Ravallon proposed an attractve and smple

More information

Brigid Mullany, Ph.D University of North Carolina, Charlotte

Brigid Mullany, Ph.D University of North Carolina, Charlotte Evaluaton And Comparson Of The Dfferent Standards Used To Defne The Postonal Accuracy And Repeatablty Of Numercally Controlled Machnng Center Axes Brgd Mullany, Ph.D Unversty of North Carolna, Charlotte

More information

LSSVM-ABC Algorithm for Stock Price prediction Osman Hegazy 1, Omar S. Soliman 2 and Mustafa Abdul Salam 3

LSSVM-ABC Algorithm for Stock Price prediction Osman Hegazy 1, Omar S. Soliman 2 and Mustafa Abdul Salam 3 LSSVM-ABC Algorthm for Stock Prce predcton Osman Hegazy 1, Omar S. Solman 2 and Mustafa Abdul Salam 3 1, 2 (Faculty of Computers and Informatcs, Caro Unversty, Egypt) 3 (Hgher echnologcal Insttute (H..I),

More information

Allocating Time and Resources in Project Management Under Uncertainty

Allocating Time and Resources in Project Management Under Uncertainty Proceedngs of the 36th Hawa Internatonal Conference on System Scences - 23 Allocatng Tme and Resources n Project Management Under Uncertanty Mark A. Turnqust School of Cvl and Envronmental Eng. Cornell

More information

Methodology to Determine Relationships between Performance Factors in Hadoop Cloud Computing Applications

Methodology to Determine Relationships between Performance Factors in Hadoop Cloud Computing Applications Methodology to Determne Relatonshps between Performance Factors n Hadoop Cloud Computng Applcatons Lus Eduardo Bautsta Vllalpando 1,2, Alan Aprl 1 and Alan Abran 1 1 Department of Software Engneerng and

More information

7.5. Present Value of an Annuity. Investigate

7.5. Present Value of an Annuity. Investigate 7.5 Present Value of an Annuty Owen and Anna are approachng retrement and are puttng ther fnances n order. They have worked hard and nvested ther earnngs so that they now have a large amount of money on

More information

Traffic State Estimation in the Traffic Management Center of Berlin

Traffic State Estimation in the Traffic Management Center of Berlin Traffc State Estmaton n the Traffc Management Center of Berln Authors: Peter Vortsch, PTV AG, Stumpfstrasse, D-763 Karlsruhe, Germany phone ++49/72/965/35, emal peter.vortsch@ptv.de Peter Möhl, PTV AG,

More information

Risk Model of Long-Term Production Scheduling in Open Pit Gold Mining

Risk Model of Long-Term Production Scheduling in Open Pit Gold Mining Rsk Model of Long-Term Producton Schedulng n Open Pt Gold Mnng R Halatchev 1 and P Lever 2 ABSTRACT Open pt gold mnng s an mportant sector of the Australan mnng ndustry. It uses large amounts of nvestments,

More information

CONSTRUCTING A SALES FORECASTING MODEL BY INTEGRATING GRA AND ELM:A CASE STUDY FOR RETAIL INDUSTRY

CONSTRUCTING A SALES FORECASTING MODEL BY INTEGRATING GRA AND ELM:A CASE STUDY FOR RETAIL INDUSTRY Internatonal Journal of Electronc Busness Management, Vol. 9, o. 2, pp. 107-121 (2011) 107 COSTRUCTIG A SALES FORECASTIG MODEL BY ITEGRATIG GRA AD ELM:A CASE STUDY FOR RETAIL IDUSTRY Fe-Long Chen and Tsung-Yn

More information