S. Chatterjee and B. Price, Regression Analysis by Example, John Wiley & Sons, 1977.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....# # , # # , # # # # # # # # , # # # # # # # # # # , # # # # # # # # , # # , # # , # # , # # , # # , # # , # # , # # , # # , # # # ] In the above formulation, let # # represent the estimate of C, and # # represent the estimate of E. Solution of the matrix equation using the pseudo inverse method [23] yields the values for the energy coefficients vector # as shown in Equation (5) such that the square error # # # # # # # # is minimized. # # # # # (5) The macro model with these energy coefficient values best fits the data acquired using the test programs. V. E XPERIMENTAL RESULTS We have ....

S. Chatterjee, A. S. Hadi, and B. Price, Regression Analysis by Examples, 3rd ed., John Wiley & Sons, 2000.

....when the values of one function are estimated through another, we have chosen an error measurement independent of the shift of one function with respect to the other. We evaluate the adequateness of each prediction of locality using a linear regression coefficient (the goodness of fit index) [4]. Consider an independent variable u representing the values of cache misses u i measured through simulation for the matrices in a group, and one dependent variable v that takes as values v i the predictions of locality for the same matrices: 1 Gamma (u i Gamma b u) u i Gamma u) ....

S. Chatterjee and B. Price. Regression Analysis by Example. John Wiley and Sons, Inc., 1991.

....our system, we will concentrate on using only the parametric methods, especially, parametric regression methods for the conversion function discovery purpose. Regression analysis is one of the most well studied method for discovering such quantitative relationships among variables from a data set [1,7]. If we view the database attributes as variables, the conversion functions we are looking for among the attributes are nothing more than regression functions. For example, the equation stk rpt.price = stock.close rate can be viewed as a regression function, where stk rpt.price is the response ....

S. Chatterjee, B. Price, Regression Analysis by Example, 2nd edn., Wiley, New York, 1991.

....We assume that the error e i has Gaussian distribution, or . Then we estimate (a m ,b m ) to maximize the probability of the error. Specifically we maximize (3.4) which is accomplished by minimizing (3.5) We can see that Eqs. 3.3) and (3.5) give the same result if is constant. The solution is [5] x y y = a x b Figure 3.1 Linear regression between sample x i and y i e i N 0 s e 2 , fe( f i e i ( i 1 = N k 1 2s e 2 y i a m x i b m ( 2 i 1 = N exp = Q m 1 s e 2 y i a m x i b m ( 2 i 1 = N = s e ....

S. Chatterjee and B. Price, Regression Analysis by Example, John Wiley & Sons, 1977.

....We assume that the error e i has Gaussian distribution, or . Then we estimate (a m ,b m ) to maximize the probability of the error. Specifically we maximize (3.4) which is accomplished by minimizing (3.5) We can see that Eqs. 3.3) and (3.5) give the same result if is constant. The solution is [5] x y y = a x b Figure 3.1 Linear regression between sample x i and y i e i N 0 s e 2 , fe( f i e i ( i 1 = N k 1 2s e 2 y i a m x i b m ( 2 i 1 = N exp = Q m 1 s e 2 y i a m x i b m ( 2 i 1 = N = s e ....

S. Chatterjee and B. Price, Regression Analysis by Example, John Wiley & Sons, 1977.

....all possible subset models and choose the best one(s) among them according to some criterion. However, evaluating all possible models may not be practically feasible when the number of variables is large. To reduce the amount of computation, two types of selection procedures have been proposed [3]: the forward selection procedure and the backward elimination procedure. The forward selection procedure starts with a model containing no variables, i.e. only a constant term, and introduces explanatory variables into the regression model one at a time. The backward elimination procedure starts ....

S. Chatterjee and B. Price. Regression Analysis by Example, 2nd Ed. John Wiley & Sons, Inc., 1991.

....all possible subset models and choose the best one(s) among them according to some criterion. However, evaluating all possible models may not be practically feasible when the number of variables is large. To reduce the amount of computation, two types of selection procedures have been proposed [2] : the forward selection procedure and the backward elimination procedure. The forward selection procedure starts with a model containing no variables, i.e. only a constant term, and introduces explanatory variables into the regression model one at a time. The backward elimination procedure ....

....1 Gammaff=2 can be found in [9] If HA is concluded for the absolute residuals and fitted values, the assumption of equal variances is violated. If the assumption of equal variances is violated, the estimates given by the corresponding regression model will not have the maximum precision [2] . Since the estimation precision requirement is not high for query optimization, the violation of this assumption can be tolerated to a certain degree. However, if the assumption of equal variances is severely violated, account should be taken of this in fitting the model. A useful tool to ....

S. Chatterjee and B. Price. Regression Analysis by Example, 2nd Ed. John Wiley & Sons, Inc., 1991.

....Figure 1: Linear Regression mapping NWS (16 MB messages) to SRB (16 MB le transfers) as predicted by RBW seem to correlate with observed le transfer behavior. 3. 1 Regression Modeling Regression modeling is in general a simple method for establishing a functional relationship among variables [8, 9, 6]. In order to achieve more accurate predictions, we considered the use of a linear regression model to address the discrepancy, exhibited in the RBW model, between the performance behavior of small NWS probes and larger le transfers. We developed two linear models that map NWS bandwidth ....

S. Chatterjee and B. Price. Regression Analysis by Example. John Wiley & Sons, Inc., 1991.

....about 20 hours on Pentium II (350MHz) PC running Matlab code. Obviously, evaluation of all possible subset models is not feasible. There are several variable selection procedures such as forward selection and backward elimination procedures, however none of them are recommended for collinear data (Chatterjee, 2000). The 24 predictors in the fouling problem are highly collinear: the condition number is 2023 (data are collinear if the condition number is greater than 100) the sum of reciprocals of the eigenvalues is 3013 which is greater than 125 times the number of predictors (data are collinear if the sum ....

Chatterjee, S., Hadi, A.S., Price, B., 2000. Regression Analysis by Examples, John Wiley & Sons, Inc.

....working on extending our model averaging strategies to include transformation selection and outlier identification. A Data for Figure 1 Table 8: Data from selected textbooks used to make Figure 1. number of page obser number of Data set Source number vations predictors Attitude Survey Chatterjee and Price (1991) 70 30 6 Equal Education Chatterjee and Price (1991) 176 70 3 Opportunity Gasoline Mileage Chatterjee and Price (1991) 261 30 10 Nuclear Power Cox and Snell (1982) 81 32 10 Crime Cox and Snell (1982) 170 47 13 Hald Draper and Smith (1981) 630 13 4 Grades Hamilton (1993) 83 118 3 Swiss Fertility ....

....to include transformation selection and outlier identification. A Data for Figure 1 Table 8: Data from selected textbooks used to make Figure 1. number of page obser number of Data set Source number vations predictors Attitude Survey Chatterjee and Price (1991) 70 30 6 Equal Education Chatterjee and Price (1991) 176 70 3 Opportunity Gasoline Mileage Chatterjee and Price (1991) 261 30 10 Nuclear Power Cox and Snell (1982) 81 32 10 Crime Cox and Snell (1982) 170 47 13 Hald Draper and Smith (1981) 630 13 4 Grades Hamilton (1993) 83 118 3 Swiss Fertility Mosteller and Tukey (1977) 550 47 5 Surgical Unit ....

[Article contains additional citation context not shown here]

Chatterjee, S. and Price, B. (1991), Regression analysis by example, 2nd edition, New York: Wiley.

....the performance behavior of remote file transfers, the complexity level of this mechanistic model might be so high that its dynamic parameterization at run time might prove infeasible. Regression modeling is a simple and lightweight method for establishing a functional relationship among variables [9, 11, 14]. In order to achieve more accurate predictions, we considered the use of a linear regression model to address the discrepancy between the performance behavior of small NWS probes and larger data transfers exhibited in the RBW model. We developed two linear models that map NWS bandwidth ....

S. Chatterjee and B. Price. Regression Analysis by Example. John Wiley & Sons, Inc., 1991.

....We propose to use regression techniques to approximate the actual attribute values of the data. Regression analysis is a statistical technique for investigating multidimensional multivariate data. It provides a conceptually simple method for establishing a functional relationships among variables [2]. We applied Least Squares Regression, which minimizes the sum of squares of differences between the observed values and the corresponding approximate values, to compute the set of coefficients of the fitting function. We refer to databases that use regression functions to model numerical data as ....

....regression model. X i can be X 2 i , 1=X i , X i X j , X i =X j , or log X i , sin X i , X 0 i , or any other transforms of X i , or combination of any of these transforms. These forms can be chosen into a regression model as the basis fitting function. Additional details can be found in [1, 2, 6, 14]. Regression Techniques Domain of Optimality Least Squares when error distribution is exactly normal Least Absolute Deviations when error distribution has heavy tails and is effective at controlling bias Huber M estimate when error distribution has heavy tails Nonparametric when error ....

S. Chatterjee and B. Price, Regression Analysis by Example, John Wiley & Sons, 1977.

....obtain improved decisions based on the gained information. The application of DA has a wide range and occurs in diverse areas where different problem formulations exist. Different algorithmic methods for DA have been suggested in the literature, as Clustering algorithms [26] regression techniques [6], Neural Networks [42] FRBSs [3] EAs [1] etc. As regards DA in the light of EAs, a representation of the information structure is considered and evolved until having an abstraction and generalization of the problem, reflected in the fitness function. For example, in [17] different approaches ....

S. Chatterjee, B. Price, Regression Analysis by Examples, John Wiley and Sons, 1991.

....the appropriate model, whereas in the field of machine learning they were created to replace the human analyst in the model selection process. 2. 1 Statistical Approaches Model selection in statistics refers to the process of estimating the relationships among the variables of a given data set (Chatterjee Price, 1977). Given one dependent variable, y, and k independent variables, x 1 ; x 2 ; x k , the goal of statistical model selection is to find a functional relationship, y = f(x 1 ; x 2 ; x k ; which explains or predicts the data. The x i are often called predictor variables because the ....

....of a matrix of weights, Omega Gamma which captures all systematic information about the disturbance process. Examination of the fit of a particular regression equation to the data requires a further assumption about the residuals: for a small data set the e i must be normally distributed (Chatterjee Price, 1977). Due to the Central Limit Theorem this is always true for large data sets. The first two assumptions, that the residual is a random variable, with mean zero and constant variance, oe 2 , follow from the assumptions about the disturbance terms (i.e, that the i ; i = 1; k are random ....

[Article contains additional citation context not shown here]

Chatterjee, S., & Price, B. (1977). Regression analysis by example. New York: John Wiley and Sons.

....Xb (4.5) minimise a cost function related to the error in the estimates. Many different linear system implementations are possible [105] and in this case the model is optimised so as to minimise the sum of the squares of the errors, E = y Gamma y) y Gamma y) 4. 6) This expression is minimised [32] by solving the system of equations known as the normal equations: X T X) b = X T y (4.7) Provided that (X T X) is non singular, b can then be found as b = X T X) Gamma1 X T y (4.8) A separate vector of regression coefficients is estimated for each output in a multivariable system, ....

S. Chatterjee and B. Price. Regression Analysis by Example. John Wiley & Sons, U.S.A., 1977.

....this paper. The least squares method (LS) fits the model (1.2) in a nonrobust way. For instance, it is possible to apply the standard calculations by processing the dummy variables in the same manner as the continuous ones, as described by Draper and Smith (1981) Montgomery and Peck (1982) and Chatterjee and Price (1977). We can also carry out an analysis of covariance, which directly assumes x p 1 to be a categorical variable. We refer to Montgomery (1991) and Edwards (1985) for more details. Unfortunately, the least squares method is very sensitive to outliers. Even a small fraction of contamination can ....

Chatterjee, S., and Price, B. (1977), Regression Analysis by Example, New York: John Wiley.

....BMA to include transformation selection and outlier identification (in work reported elsewhere Hoeting et al. 1995, 1996) Appendix A: Data for Figure 1 Data from selected textbooks used to make Figure 1. number of page obser number of Data set Source number vations predictors Attitude Survey Chatterjee and Price (1991) 70 30 6 Equal Education Chatterjee and Price (1991) 176 70 3 Opportunity Gasoline Mileage Chatterjee and Price (1991) 261 30 10 Nuclear Power Cox and Snell (1982) 81 32 10 Crime Cox and Snell (1982) 170 47 13 Hald Draper and Smith (1981) 630 13 4 Grades Hamilton (1993) 83 118 3 Swiss Fertility ....

....identification (in work reported elsewhere Hoeting et al. 1995, 1996) Appendix A: Data for Figure 1 Data from selected textbooks used to make Figure 1. number of page obser number of Data set Source number vations predictors Attitude Survey Chatterjee and Price (1991) 70 30 6 Equal Education Chatterjee and Price (1991) 176 70 3 Opportunity Gasoline Mileage Chatterjee and Price (1991) 261 30 10 Nuclear Power Cox and Snell (1982) 81 32 10 Crime Cox and Snell (1982) 170 47 13 Hald Draper and Smith (1981) 630 13 4 Grades Hamilton (1993) 83 118 3 Swiss Fertility Mosteller and Tukey (1977) 550 47 5 Surgical Unit ....

[Article contains additional citation context not shown here]

Chatterjee, S. and Price, B. (1991), Regression Analysis by Example, 2nd edition, New York: Wiley.

No context found.

S. Chatterjee and B. Price, Regression Analysis by Example, John Wiley & Sons, 1977.

No context found.

S. Chatterjee, A. S. Hadi, and B. Price, Regression Analysis by Examples, 3rd ed., John Wiley & Sons, 2000.

No context found.

S. Chatterjee, A. S. Hadi, and B. Price, Regression Analysis by Examples, 3rd ed. New York: Wiley, 2000.

No context found.

S. Chatterjee, A. S. Hadi, and B. Price. Regression Analysis by Example, Third Edition. John Wiley & Sons, Inc, 2000. ISBN 0-471-31946-5.

No context found.

S. Chatterjee, A. S. Hadi, and B. Price. Regression Analysis by Example, Third Edition. John Wiley & Sons, Inc., 2000. ISBN 0-471-31946-5.

No context found.

S Chatterjee and B Price, Regression Analysis by Example, Wiley 1977 (SF 2.5 CHA).

No context found.

Chatterjee, S., Price, B., 1991. Regression Analysis by Example. 2nd ed., John Wiley & Sons.

No context found.

S Chatterjee and B Price, Regression Analysis by Example, Wiley 1977 (SF 2.5 CHA).

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC