EVALUATION AND RESEARCH OF INCUBATOR S HATCH ABILITY OF SMALL AND MEDIUM-SIZED ENTERPRISE BASED ON MULTICLASS ANALYSIS

Transcription

1 Journal of Theoretical and Alied Inforation Technology EVALUATION AND RESEARCH O INCUBATOR S HATCH ABILITY O SMALL AND MEDIUM-SIZED ENTERPRISE BASED ON MULTICLASS ANALYSIS XUE WU Chongqing University of Science and Technology College of Business Adinistration, Chongqing,400020,China ABSTRACT With the continuous develoent of business incubator nowadays, the eisting evaluation ethod of enterrise incubator s hatch ability has been unable to eet the needs of sall and ediu-sized enterrises. This aer rooses a sall and ediu-sized enterrise incubator s hatch ability evaluation odel based on ulticlass ethod, this odel introduced the rincial coonent analysis and factor analysis, first, it selects sall and ediu-sized enterrise incubator s hatch ability evaluation inde, then it akes data conversion to these indicators, then it carries out the variable correlation analysis, and finally do the hatch ability evaluation. The eeriental results show that the sall and ediu-sized enterrise incubator s hatch ability evaluation odel based on ulticlass can evaluate effectively sall and ediusized enterrise incubator s ability, that is to say this odel is ractical and effective. Keywords: Princial Coonent Analysis, actor Analysis Method, Sall and Mediu-sized Enterrise Incubator, Hatchability Evaluation 1. INTRODUCTION Business incubator is a new tye of social and econoic organization, it reduces venture business risk and business cost, enhances the enterrises survival rate and the success rate through roviding the research, develoing, and anaging site, the sharing facilities like counication, network and office tools, and by roviding the systeatic training and consulting, and the suorting of olitical, financial, legal and arket rootion [1]. However, with the continuous develoent of our country s enterrise incubator, it also aears soe robles in its oeration, such as the lag of basic services ability, the low of incubation effect, the low quality of incubator anageent ersonnel, these all affect the enterrise s ability of incubator hatch ability (2-3). Therefore, when our eyes are focusing on the achieveent of our country s enterrise incubator over twenty years, we also should notice the robles that we have [4-6]. or the research of enterrise incubator s hatch ability evaluation, such as Chan and Lau ut forward the science and technology business incubator related assessent fraework by using the organization theory and cluster theory, the fraework including nine evaluation criteria: resource collection, resource sharing, consulting services, ublic iage, the network latfor, geograhical advantages, cost advantage, and financing suort (7), In Mo Guang Zheng study of higher school social service evaluation, he ut forward a enterrise incubator hatchability inde syste in university science and technology ark, constructing the incubator s hard and soft environent condition, incubator s financing ability, anageent ability, talent gathered ability, arketing ability, and the key factors, which consists of 17 indicators [8]. As for foreign study, foreign scholars started earlier the study of the assessent and research of business incubator, who has ade great achieveents, and latter they also use a quantitative ethod to evaluate and analysis, which has the certain scientific, but lack of syste related theory, and no secialized hatch ability analysis, what s ore, the research according to the actual situation of the country s enterrise incubator s develoent, the research results has a certain significance to the evaluation theory of enterrise incubator hatchability, but we need ore study about our country enterrise incubator s hatch ability s research and evaluation. Based on the evaluation and research status of sall and ediu-sized enterrise incubator s hatch ability, this aer ut forward a kind of sall and ediu-sized enterrise incubator s hatch ability evaluation based on ulticlass analysis, and using rincial coonent analysis and factor analysis to go the corehensive evaluation. 1527

2 Journal of Theoretical and Alied Inforation Technology 2. PINCIPLE COMPONENT ANALYSIS Princial coonent analysis is a kind of coonly used ultivariate statistical ethod, which has been widely used in the social econoy, enterrise anageent and geological, biocheical, haraceutical, and other fields, such as clothing finalize the design, enterrise's econoic benefit analysis, and the ai aking of new develoent roect [9]. The target of rincial coonent analysis is to silify the ultivariate statistics syste. That is to say, we carry out the diensional reduction treatent to the high diension variable sace under the rincile of least losing the data inforation [10]. Assuing the assessent questions has s inde 1, 2,...,,these s inde indicate the natures of all the assessent obects, thus the observed s inde value of every obect is a sale value, which is a vector of s diension. If there are n s obectives, there are n s vectors of s diension, showing in atri X : X = (1) n 1 n2... n If we take s inde 1, 2,..., as the rando variables,their eecting value and covariance atri are 1 µ 1 2 µ 2 E = E = (2) µ v11 v12... v1 v21 v22... v 2 V = ( vi ) = ( Cov( i, )) = v 1 v2... v vi is the variance of the variable of i,thus the change status of these s variables can also be indicated by i= 1 v i. The roble now is whether we can find a are of new variables y, y,..., y ( < ),that can ake 1 2 (3) L y = L, = 1,..., (4) γ = { L: LL = 1}, γ is called the collection of regulated vectors of s diension. Which satisfy: L (1) y 1 is in all the siilar fors to y= reaching the aiu variance L γ ; (2) y 2 is in all the siilar fors to y= not related with y1 and reaching the aiu variance L γ ; (3) y k is in all the siilar fors to y= not related with y1,..., yk 1and reaching the aiu variance L γ, k = 3,4,...,. This roble is ust to solve a Var( y) = Var( L ) = L VL,and LL 1 Using the Lagrange ultilier ethod to establish the obective function Q LVL λ( LL 1) L L = (5) = (6) Q Thus = 2VL 2λL L Q and = 0,then we have VL = λl, L LL = 1. This indicates that L is the facture vector of atri V,and under the above forula, we have Var( y) = L VL = λl L = λ (7) Naely characteristic root λ is variance of variable y. So only to solve the feature vector L corresonding to the biggest characteristic root, the ain ingredients can work out. So the found new variables y 1, y 2,..., y are all the linear cobination of the original variable 1, 2,...,. The coordinate syste constructed by y1, y2,..., y is got by the translation and orthogonal rotation of the original coordinate syste; we say the sace constructed by y1, y2,..., y is the ain hyerlane of s diensions. 1528

3 Journal of Theoretical and Alied Inforation Technology 3. ACTOR ANALYSIS METHOD 3.1 actor Analysis Model actor analysis using a sall aount of corehensive inde (ain factor) instead of ultile riary indees, the got ain factor is for the linear cobination of the original inde. Assuing there are s observation variables 1, 2,...,,ut these into standardization, aking the ean of the variable as 0 after standardization, and the variance is 1. Recording the original ublic factor s variables as f1, f2,..., f,and the ublic factors after the standardization are 1, 2,..., ( < ), s ublic factors that can not resented are called secial factors, arked as ε1, ε2,..., ε. And the factor analysis odel is 1 = a111 + a a1 + ε1 2 = a211 + a a2 + ε 2 (8)... = a 11 + a a + ε In which, a is the factor loading, the greater i the absolute value of a i is( ai 1), i deends greater on,all the eleents a i coosing the factor loading atri A. 3.2 Assessent Stes (1)Constructing the original atri X according to the selected inde syste and the inde data, and carry out the standardization to the atri X, and get the standardized atri Z ; (2) Using Z to calculate correlation coefficient atri R, order R λl = 0, and solving the characteristic value, the contribution rate and cuulative contribution rate of R, under the rincial of the ain coonent characteristic root not less than 1 or ain coonent s variance cuulative contribution rate not less than 85%, thus we can deterine the nuber of ain factor (set < ); (3) Calculation feature vector and initial factor loading atri A, using the ethod of regression to estiate factor score, taking the roortion of each factor s variance contribution rate occuies total variance factor contribution rate as weight to go the weighted suary, finally get factor s corehensive analysis odel λ1 λ2 λ = λ λ λ i i i i= 1 i= 1 i= 1 (9) (4) If factor loading quantity is averagely, and difficult to discriinate which indicator contacts closely with a certain factor, then using the aiu variance ethod to do orthogonal rotation of the initial factor, aking the contribution of the ublic factors diserse the better, and the original inde have the large load only in a ublic factor. In this way we can give the ain factor reasonable econoic eaning to analyze the researching roble. (5) According to the sale factor score to do the corehensive analysis and evaluation, the higher the score shows the better roerties of the sale. 4. EVALUATION MODEL O INCUBATOR HATCHABILITY 4.1 Inde Choosing When evaluating the sall and ediu-sized enterrise s incubator hatchability, we ust first select evaluation inde. The quantitative indicators selected by this article show as follows: sall and ediu-sized enterrise incubator site area 1, sall and ediusized enterrise graduation rates 2, the nuber of eloyees of enterrise in hatch 3,the total incoe of the enterrises in hatch 4, the ta ayent of the enterrises in hatch 5, staff s high quality structure 6, the nuber of new hatched enterrises 7, the net value of fied assets by the end of the year 8, the total fund obtained by sall and ediu-sized enterrise incubator Data Conversion In order to ake it ossible for the coarison between the indees, and eliinate the ossible influence on the evaluation result caused by observation and diension differences, thus we need to carry out the standardization rocessing to the inde data of incubator hatchability, using SPSS software to start the original data standardization rocess. The secific coutation forula is as follows: Assuing the original data atri is = ( )( i = 1,2,..., ; = 1,2,..., N) (10) i 1529

4 Journal of Theoretical and Alied Inforation Technology In which i is the s inde value of i s incubator sale, N is the evaluation inde nuber. Then carry out the Diensionless rocessing according to the forula below: ( i µ ) Zi = (11) S Zi In the forula(6), is the standard value of i µ data, is the average value of inde S, is the standard error of inde. After the standardization rocessing of the i sale observed data,we can get the average value 0 of the standard value Zi ( = 1,2,..., ),variance Correlation Analysis of The Variance Because one of the ain task of the factor analysis is to concentrate the original variables, which is to ut forward the overlas inforation of the original variable and integrate into the factor, and finally realize the urose of reduce the nuber of variables, for this rogress, it requires a strong correlation between the original variable, or if the original variables are indeendent, it cannot be concentrated, and also has no need for factor analysis. Checking whether there is relationshi between variables, we usually can use the ethod of calculating sile correlation coefficient atri, calculating the inverse iage correlation atri (Anti-iage correlation atri), and the Bartlett Shericity insection (Bartlett of Shericity), KMO (Kaiser - Meyer - Olkin) insection ethods to study the relationshi between the original variables. This aer takes KMO and Bartlett sherical to analyze the correlation of variables. 5. INCUBATOR S HATCH ABILITY EVALUATION O SMALL AND MEDIUM-SIZED ENTERPRISE This aer selects 134 sall business incubators as a sale for incubator s hatch ability evaluation. irst take KMO and Bartlett sherical to analyze the correlation of variables. Table 1 shows the test results, the first behavior test the correlation KMO statistics between variables, its value is > 0.5, shows that factor analysis effect is good; The second line hyothesis the test chi-square value is , the associated robability is 0.000, less than significance level 0.05, obviously, the correlation coefficient atri is not unit array, and it has strong correlation between variances. Therefore, the original data selected by this aer is roer to use the factor analysis. And through the various factor etent sizes, the ain factors of these nine factors are first three factors. Table 1: KMO and Bartlett s Test Kaiser-Meyer-Olkin Measure of Saling Adequacy Bartlett s Test of Aro.Chi-Sqiare Shericity df 36 Sig Then use Varia ethod to carry on the factor rotation of first three factors, the results show as table 2 and table 3, in which the table 2 is the factor loading atri before rotating, table 3 is the factor loading atri after rotating. Table 2: Coonent Matri(a) Coonent Zscore( 1) Zscore( 2) Zscore( 3) Zscore( 4) Zscore( 5) Zscore( 6) Zscore( 7) Zscore( 8) Zscore( 9) Table 3:Rotated Coonent Matri(A) Coonent Zscore( 1) Zscore( 2) Zscore( 3) Zscore( 4) Zscore( 5) Zscore( 6) Zscore( 7) Zscore( 8) Zscore( 9)

5 Journal of Theoretical and Alied Inforation Technology We can see it out the constructive relation between factors fro the factors after the rotation of Table 3. Aong the, the first factor 1 ainly reflects the enterrise ersonnel nuber in the hatch, enterrise incoe in hatch, the new hatch enterrise nuber and the ay taes, and the first factor ainly anifests the enterrise incubator's innovation ability and their incubation scale. The second factor 2 ainly reflects the incubation fund and staff s high quality structure, ainly anifests the enterrise incubator s anageent and service ability, the suerior the quality of the staff structure is, the stronger of its anageent ability is, but as for incubation fund, it has the influence to the self develoent of enterrise incubator. The third factor 3 reflects the enterrise graduation rates and net value of fied assets by the end of the year, which ainly anifests the enterrise incubator s self develoent ability and incubation efficiency. The above three factors covers 8 of the 9 indicators, who is good for reresentativeness and elanatory. We can get fro Table 3 the roortion of each factor etracts each inde inforation. Thus, we can see that: In the first rincial coonent 1, eloyees of enterrise in hatch, total incoe of the enterrise in hatch, the ta ayent and new hatch enterrise are with high load, which elains this several variables has a strong relationshi with this factor, and the higher the inde value is, the stronger the relationshi is between the variables and factor, or it becoes ore distance, fro that, we can see that in the variables who reflect the enterrise incubator innovation ability, the iortant influence factors is the total incoe of enterrise in hatch. In the second rincial coonents 2, the structure of staff quality and incubation fund have high load, so it deonstrates that the two variables have a closely relationshi with the factors, and incubation fund has the largest contribution for ain coonents, which reflects in the incubator self-develoent ability, the iortant factors are the incubation fund. In the third rincial coonent 3, enterrise graduation rates and net value of fied assets by the end of the year are with high load, so it deonstrates that the two variables have the closely relationshi with the factors, and enterrise graduation rate has the largest contribution for ain coonents 3, which reflects the enterrise graduation rate is also an iortant factor which influences the enterrise incubator innovation ability. 6. CONCLUSION This aer evaluates the incubate ability of SME incubator with the rincile coonent analysis and factor analysis, and akes the eale evaluation to the incubate ability of 134 SME incubators in our country. Results show that, the roosed evaluation ethod is ractical and effective; it is worth for sreading use. ACKNOWLEDGEMENTS Collaborative incubation roble research of Sciencifical and icro enterrise. Ite nuber:cstc 2012c-rka0063 REERENCES: [1] Q. Y. Yu, Q. J. Wang, J. Du. "Evaluation of enterrise's incubator's hatch ability," Science and technology, Vol. 8, No. 20, 2001, [2] X. Y. Cao. "Evaluation and research of enterrise s incubator's hatch ability," Scientifically rogress and strategy, Vol. 18, No. 6, 2001, [3] M. K. Shen. "The eirical research of factor analysis ethod on the choice of regional and strategic eerging industry," Enterrise econoy, Vol. 18, No. 9, 2012, [4] Q. S. Zhu, G. C. Meng, Y. Ge. "The online video shot boundary detection based on increental rincial coonent analysis," Couter engineering and science, Vol. 34, No. 1, 2012, [5] X. N. Pang, L. Wang. "Indicator syste construction of circular econoic corehensive evaluation Based on the rincial coonent analysis ethod," Statistics and decision, Vol. 3, No. 14, 2012, [6] M. John, S. P. Yu. "Corehensive evaluation of Xiniang urbanization develoent level Based on the rincial coonent analysis and cluster analysis," Technology anageent and research, Vol. 32, No. 21, 2012, [7] Y. J. Pu, Q. Zhang, S. Huang. "Analysis of efficiency and the influence factors of econoic 1531

6 Journal of Theoretical and Alied Inforation Technology develoent in the western region - based on Malquisit inde and satial econoetric odel cobination," Systeatic engineering, Vol. 30, No. 9, 2012, [8] A. P. Zhou, G. Cheng. "Analysis of network behavior characteristics and the influence factor Based on the eak flow," Journal of counication, Vol. 33, No. 10, 2012, [9]. J. Ren. "Henan industrial advantage research based on the rincial coonent analysis," Scientific anageent and research, Vol. 32, No. 18, 2012, [10] H. Y. an, D. Z. Liu, Y. H. Ma. "Three hase integration odel of factor s eleent analysis," Psychological scientifically rogress, Vol. 20, No. 10, 2012,