ࡱ> xG pvbjbjَ \Zc]@@@@7zzz. B  6666666$A95;6  6e%@@z. e%e%e% @z.6hHh@@@@ 6e%Je%*446 .T > rN6>  MAR 5621 Advanced Statistical Techniques Summer 2003 Dr. Larry Winner Chapter 11 Simple linear regression Types of Regression Models (Sec. 11-1) Linear Regression:  EMBED Equation.COEE2   EMBED Equation.COEE2  - Outcome of Dependent Variable (response) for ith experimental/sampling unit  EMBED Equation.COEE2  - Level of the Independent (predictor) variable for ith experimental/sampling unit  EMBED Equation.COEE2  - Linear (systematic) relation between Yi and Xi (aka conditional mean)  EMBED Equation.COEE2  - Mean of Y when X=0 (Y-intercept)  EMBED Equation.COEE2  - Change in mean of Y when X increases by 1 (slope)  EMBED Equation.COEE2  - Random error term Note that  EMBED Equation.COEE2  and  EMBED Equation.COEE2  are unknown parameters. We estimate them by the least squares method. Polynomial (Nonlinear) Regression:  EMBED Equation.COEE2  This model allows for a curvilinear (as opposed to straight line) relation. Both linear and polynomial regression are susceptible to problems when predictions of Y are made outside the range of the X values used to fit the model. This is referred to as extrapolation. Least Squares Estimation (Sec. 11-2) Obtain a sample of n pairs (X1,Y1)(Xn,Yn). Plot the Y values on the vertical (up/down) axis versus their corresponding X values on the horizontal (left/right) axis. Choose the line  EMBED Equation.COEE2  that minimizes the sum of squared vertical distances from observed values (Yi) to their fitted values ( EMBED Equation.COEE2 ) Note:  EMBED Equation.COEE2  b0 is the Y-intercept for the estimated regression equation b1 is the slope of the estimated regression equation Measures of Variation (Sec. 11-3) Sums of Squares Total sum of squares = Regression sum of squares + Error sum of squares Total variation = Explained variation + Unexplained variation Total sum of squares (Total Variation):  EMBED Equation.COEE2  Regression sum of squares (Explained Variation):  EMBED Equation.COEE2  Error sum of squares (Unexplained Variation):  EMBED Equation.COEE2  Coefficients of Determination and Correlation Coefficient of Determination Proportion of variation in Y explained by the regression on X  EMBED Equation.COEE2  Coefficient of Correlation Measure of the direction and strength of the linear association between Y and X  EMBED Equation.COEE2  Standard Error of the Estimate (Residual Standard Deviation) Estimated standard deviation of data ( EMBED Equation.COEE2 )  EMBED Equation.COEE2  Model Assumptions (Sec. 11-4) Normally distributed errors Heteroscedasticity (constant error variance for Y at all levels of X) Independent errors (usually checked when data collected over time or space) Residual Analysis (Sec. 11-5) Residuals:  EMBED Equation.COEE2  Plots (see prototype plots in book and in class): Plot of ei vs  EMBED Equation.COEE2  Can be used to check for linear relation, constant variance If relation is nonlinear, U-shaped pattern appears If error variance is non constant, funnel shaped pattern appears If assumptions are met, random cloud of points appears Plot of ei vs Xi Can be used to check for linear relation, constant variance If relation is nonlinear, U-shaped pattern appears If error variance is non constant, funnel shaped pattern appears If assumptions are met, random cloud of points appears Plot of ei vs i Can be used to check for independence when collected over time (see next section) Histogram of ei If distribution is normal, histogram of residuals will be mound-shaped, around 0 Measuring Autocorrelation Durbin-Watson Test (Sec. 11-6) Plot residuals versus time order If errors are independent there will be no pattern (random cloud centered at 0) If not independent (dependent), expect errors to be close together over time (distinct curved form, centered at 0). Durbin-Watson Test H0: Errors are independent (no autocorrelation among residuals) HA: Errors are dependent (Positive autocorrelation among residuals) Test Statistic:  EMBED Equation.COEE2  Decision Rule (Values of  EMBED Equation.COEE2  and  EMBED Equation.COEE2  are given in Table E.10): If  EMBED Equation.COEE2  then conclude H0 (independent errors) where k is the number of independent variables (k=1 for simple regression) If  EMBED Equation.COEE2  then conclude HA (dependent errors) where k is the number of independent variables (k=1 for simple regression) If  EMBED Equation.COEE2  then we withhold judgment (possibly need a longer series) NOTE: This is a test that you want to include in favor of the null hypothesis. If you reject the null in this test, a more complex model needs to be fit (will be discussed in Chapter 13). Inferences Concerning the Slope (Sec. 11-7) t-test Test used to determine whether the population based slope parameter ( EMBED Equation.COEE2 ) is equal to a pre-determined value (often, but not necessarily 0). Tests can be one-sided (pre-determined direction) or two-sided (either direction). 2-sided t-test:  EMBED Equation.COEE2  1-sided t-test (Upper-tail, reverse signs for lower tail):  EMBED Equation.COEE2  F-test (based on k independent variables) A test based directly on sum of squares that tests the specific hypotheses of whether the slope parameter is 0 (2-sided). The book describes the general case of k predictor variables, for simple linear regression, k=1.  Analysis of Variance (based on k Predictor Variables) Source df Sum of Squares Mean Square F Regression k SSR MSR=SSR/k Fobs=MSR/MSE Error n-k-1 SSE MSE=SSE/(n-k-1) --- Total n-1 SST --- --- (1-a)100% Confidence Interval for the slope parameter, b1  If entire interval is positive, conclude b1>0 (Positive association) If interval contains 0, conclude (do not reject) b1=0 (No association) If entire interval is negative, conclude b1<0 (Negative association) Estimation of Mean and Prediction of Individual Outcomes (Sec. 11-8) Estimating the Population Mean Response when X=Xi Parameter:  EMBED Equation.COEE2  Point Estimate:  EMBED Equation.COEE2  (1-a)100% Confidence Interval for  EMBED Equation.COEE2 :  EMBED Equation.COEE2  Predicting a Future Outcome when X=Xi Future Outcome:  EMBED Equation.COEE2  Point Prediction:  EMBED Equation.COEE2  (1-a)100% Prediction Interval for  EMBED Equation.COEE2   EMBED Equation.COEE2  Problem Areas in Regression Analysis (Sec. 11-9) Awareness of model assumptions Knowing methods of checking model assumptions Alternative methods when model assumptions are not met Not understanding the theoretical background of the problem The most important means of identifying problems with model assumptions is based on plots of data and residuals. Often transformations of the dependent and/or independent variables can solve the problems. Other, more complex methods may need to be used. See Section 11-5 for plots used to check assumptions and Exhibit 11.3 on pages 555-556 of text. Computational Formulas for Simple Linear Regression (Sec. 11-10) Normal Equations (Based on Minimizing SSE by Calculus)  Computational Formula for the Slope, b1  Computational Formula for the Y-intercept, b0  Computational Formula for the Total Sum of Squares, SST  Computational Formula for the Regression Sum of Squares, SSR  Computational Formula for Error Sum of Squares, SSE  Computational Formula for the Standard Error of the Slope,  EMBED Equation.COEE2   Chapter 12 mULTIple Linear regression Multiple Regression Model with k Independent Variables (Sec. 12-1) General Case: We have k variables that we control, or know in advance of outcome, that are used to predict Y, the response (dependent variable). The k independent variables are labeled X1, X2,...,Xk. The levels of these variables for the ith case are labeled X1i ,..., Xki . Note that simple linear regression is a special case where k=1, thus the methods used are just basic extensions of what we have previously done. Yi = b0 + b1X1i + ... + bkXki + ei where bj is the change in mean for Y when variable Xj increases by 1 unit, while holding the k-1 remaining independent variables constant (partial regression coefficient). This is also referred to as the slope of Y with variable Xj holding the other predictors constant. Least Squares Fitted (Prediction) Equation (obtained by minimizing SSE): Coefficient of Multiple Determination Proportion of variation in Y explained by the regression on the k independent variables.  Adjusted-R2 Used to compare models with different sets of independent variables in terms of predictive capabilities. Penalizes models with unnecessary or redundant predictors.  Residual Analysis for Multiple Regression (Sec. 12-2) Very similar to case for Simple Regression. Only difference is to plot residuals versus each independent variable. Similar interpretations of plots. A review of plots and interpretations: Residuals:  EMBED Equation.COEE2  Plots (see prototype plots in book and in class): Plot of ei vs  EMBED Equation.COEE2  Can be used to check for linear relation, constant variance If relation is nonlinear, U-shaped pattern appears If error variance is non constant, funnel shaped pattern appears If assumptions are met, random cloud of points appears Plot of ei vs Xji for each j Can be used to check for linear relation with respect to Xj If relation is nonlinear, U-shaped pattern appears If assumptions are met, random cloud of points appears Plot of ei vs i Can be used to check for independence when collected over time If errors are dependent, smooth pattern will appear If errors are independent, random cloud of points appears Histogram of ei If distribution is normal, histogram of residuals will be mound-shaped, around 0 F-test for the Overall Model (Sec. 12-3) Used to test whether any of the independent variables are linearly associated with Y Analysis of Variance Total Sum of Squares and df:  EMBED Equation.COEE2  Regression Sum of Squares:  EMBED Equation.COEE2  Error Sum of Squares:  EMBED Equation.COEE2  Source df SS MS F Regression k SSR MSR=SSR/k Fobs=MSR/MSE Error n-k-1 SSE MSE=SSE/(n-k-1) --- Total n-1 SST --- --- F-test for Overall Model H0: b1 = ... = bk = 0 (Y is not linearly associated with any of the independent variables) HA: Not all bj = 0 (At least one of the independent variables is associated with Y ) TS:  EMBED Equation.COEE2  RR:  EMBED Equation.COEE2  P-value: Area in the F-distribution to the right of Fobs Inferences Concerning Individual Regression Coefficients (Sec. 12-4) Used to test or estimate the slope of Y with respect to Xj, after controlling for all other predictor variables. t-test for bj H0: bj = bj0 (often, but not necessarily 0) HA: bj  EMBED Equation.COEE2 bj0 TS:  EMBED Equation.COEE2  RR:  EMBED Equation.COEE2  P-value: Twice the area in the t-distribution to the right of | tobs | (1-a)100% Confidence Interval for bj  EMBED Equation.COEE2  If entire interval is positive, conclude bj>0 (Positive association) If interval contains 0, conclude (do not reject) bj=0 (No association) If entire interval is negative, conclude bj<0 (Negative association) Testing Portions of the Model (Sec. 12-5, But Different Notation) Consider 2 blocks of independent varibles: Block A containing k-q independent variables Block B containing q independent variables We wish to test whether the set of independent variables in block B can be dropped from a model containing the set of independent variables. That is, the variables in block B add no predictive value to the variables in block A. Notation: SSR(A&B) is the regression sum of squares for model containing both blocks SSR(A) is the regression sum of squares for model containing block A MSE(A&B) is the error mean square for model containing both blocks k is the number of predictors in the model containing both blocks q is the number of predictors in block B H0: The b s in block B are all 0, given that the variables in block A have been included in model HA: Not all b s in block B are 0, given that the variables in block A have been included in model TS:  EMBED Equation.COEE2  RR:  EMBED Equation.COEE2  P-value: Area in Fdistribution to the right of Fobs Coefficient of Partial Determination (Notation differs from text) Measures the fraction of variation in Y explained by Xj, that is not explained by other independent variables.  EMBED Equation.COEE2   EMBED Equation.COEE2   EMBED Equation.COEE2  Note that since labels are arbitrary, these can be constructed for any of the independent variables. Quadratic Regression Models (Sec. 12-6) When the relation between Y and X is not linear, it can often be approximated by a quadratic model. Population Model:  EMBED Equation.COEE2  Fitted Model:  EMBED Equation.COEE2  Testing for any association: H0: b1=b2=0 HA: b1 and/or b2  EMBED Equation.COEE2 0 (F-test) Testing for a quadratic effect: H0: b2=0 HA: b2 EMBED Equation.COEE2 0 (t-test) Models with Dummy Variables (Sec. 12-7) Dummy variables are used to include categorical variables in the model. If the variable has m levels, we include m-1 dummy variables, The simplest case is binary variables with 2 levels, such as gender. The dummy variable takes on the value 1 if the characteristic of interest is present, 0 if it is absent. Model with no interaction Consider a model with one numeric independent variable (X1) and one dummy variable (X2). Then the model for the two groups (characteristic present and absent) have the following relationships with Y:  Characteristic Present (X2i = 1):  EMBED Equation.COEE2  Characteristic Absent (X2i=0):  EMBED Equation.COEE2  Note that the two groups have different Y-intercepts, but the slopes with respect to X1 are equal. Model with interaction This model allows the slope of Y with respect to X1 to be different for the two groups with respect to the categorical variable:  X2i=1:  EMBED Equation.COEE2  X2i=0:  EMBED Equation.COEE2  Note that the two groups have different Y-intercepts and slopes. More complex models can be fit with multiple numeric predictors and dummy variables. See examples. Transformations to Meet Model Assumptions of Regression (Sec. 12-8) There are a wide range of possibilities of transformations to improve model when assumptions on errors are violated. We will consider specifically two types of logarithmic transformations that work well in a wide range of problems. Other transformations can be useful in very specific situations. Logarithmic Transformations These transformations can often fix problems with nonlinearity, unequal variances, and non-normality of errors. We will consider base 10 logarithms, and natural logarithms.  Case 1: Multiplicative Model (with multiplicative nonnegative errors)  EMBED Equation.COEE2  Taking base 10 logarithms of both sides gives model:  EMBED Equation.COEE2  Procedure: First, take base 10 logarithms of Y and X1 and X2, and fit the linear regression model on the transformed variables. The relationship will be linear for the transformed variables. Case 2: Exponential Model (with multiplicative nonnegative errors)  Taking natural logarithms of the Y values gives the model:  EMBED Equation.COEE2  Multicollinearity (Sec. 12-9) Problem: In observational studies and models with polynomial terms, the independent variables may be highly correlated among themselves. Two classes of problems arise. Intuitively, the variables explain the same part of the random variation in Y. This causes the partial regression coefficients to not appear significant for individual terms (since they are explaining the same variation). Mathematically, the standard errors of estimates are inflated, causing wider confidence intervals for individual parameters, and smaller t-statistics for testing whether individual partial regression coefficients are 0. Variance Inflation Factor (VIF)  where  EMBED Equation.COEE2 is is the coefficient of determination when variable Xj is regressed on the j-1 remaining independent variables. There is a VIF for each independent variable. A variable is considered to be problematic if its VIF is 10.0 or larger. When multicollinearity is present, theory and common sense should be used to choose the best variables to keep in the model. Complex methods such as principal components regression and ridge regression have been developed, we will not pursue them here, as they dont help in terms of the explanation of which independent variables are associated and cause changes in Y. Model Building (Sec. 12-10) A major goal of regression is to fit as simple of model as possible, in terms of the number of predictor (independent) variables, that still explains variation in Y. That is, we wish to choose a parsimonious model (think of Ockhams razor). We consider two methods. Stepwise Regression This is a computational method that uses the following algorithm: Choose a significance level (P-value) to enter the model (SLE) and a significance level to stay in the model (SLS). Some computer packages require that SLE < SLS. Fit all k simple regression models, choose the independent variable with the largest t-statistic (smallest P-value). Check to make sure the P-value is less than SLE. If yes, keep this predictor in model, otherwise stop (no model works). Fit all k-1 variable models with two variables (one of which is the winning variable from step 2). Choose the model with the new variable having the lowest P-value. Check to determine that this new variable has P-value less than SLE. If no, then stop and keep the model with only the winning variable from step 2. If yes, then check that the variable from step 2 is still significant (its P-value is less than SLS), if not drop the variable from step 2. Fit all k-2 variable models with the winners from steps 2 and 3. Follow the process described in step 3, where both variables in model from steps 2 and 3 are checked against SLS. Continue until no new terms can enter model by not meeting SLE criteria. Best Subsets (Mallows Cp Statistic) Consider a situation with T-1 potential independent variables (candidates). There are 2T-1 possible models containing subsets of the T-1 candidates. Note that the full model containing all candidates actually has T parameters (including the intercept term). Fit the full model and compute  EMBED Equation.COEE2 , the coefficient of determination For each model with k predictors, compute  EMBED Equation.COEE2  Compute  EMBED Equation.COEE2  Choose model with smallest k so that CP is is close to or below k+1 Chapter 13 Time series & Forecasting Classical Multiplicative Model (Sec. 13-2) Time series can generally be decomposed into components representing overall level, linear trend, cyclical patterns, seasonal shifts, and random irregularities. Series that are measured at the annual level, cannot demonstrate the seasonal patterns. In the multiplicative model, effects of each component enter in a multiplicative manner. Multiplicative Model Observed at Annual Level  EMBED Equation.COEE2  where the components are trend, cycle, and irregularity effects Multiplicative Model Observed at Seasonal Level  EMBED Equation.COEE2  where the components are trend, cycle, seasonal, and irregularity effects Smoothing a Time Series (Sec. 13-3) Time series data are usually smoothed to reduce the impact of the random irregularities in individual data points. Moving Averages Centered moving averages are averages of the value at the current point and neighboring values (above and below the point). Due to the centering, they are typically odd numbered. The 3- and 5-period moving averages at time point i would be:  Clearly, their cant be moving averages at the lower and upper end of the series. Exponential Smoothing This method uses all previous data in a series to smooth it, with exponentially decreasing weights on observations further in the past:  The exponentially smoothed value for a current time point can be used as forecast for the next time point.  EMBED Equation.COEE2  Least Squares Trend Fitting and Forecasting (Sec. 13-4) We consider three trend models: linear, quadratic, and exponential. In each model the independent variable Xi is the time period of observation i. If the first observation is period 1, then Xi = i. Linear Trend Model Population Model:  EMBED Equation.COEE2  Fitted Forecasting Equation:  EMBED Equation.COEE2  Quadratic Trend Model Population Model:  EMBED Equation.COEE2  Fitted Forecasting Equation:  EMBED Equation.COEE2  Exponential Trend Model Population Model:  EMBED Equation.COEE2  Fitted Forecasting Equation:  EMBED Equation.COEE2  This exponential models forecasting equation is obtained by first fitting a simple linear regression of the logarithm of Y on X, then putting the least squares estimates of the Y-intercept and the slope into the second equation. Model Selection Based on Differences First Differences:  EMBED Equation.COEE2  Second Differences:  EMBED Equation.COEE2  Percent Differences:  EMBED Equation.COEE2  If first differences are approximately constant, then the linear trend model is appropriate. If the second differences are approximately constant, then the quadratic trend model is appropriate. If percent differences are approximately constant, then the exponential trend model is appropriate. Autoregressive Model for Trend Fit and Forecasting (Sec. 13-5) Autoregressive models allow for the possibility that the mean response is a linear function of previous response(s). 1st Order Model Population Model:  EMBED Equation.COEE2  Ftted Equation:  EMBED Equation.COEE2  Pth Order Model Population Model:  EMBED Equation.COEE2  Ftted Equation:  EMBED Equation.COEE2  Significance Test for Highest Order Term  This test can be used repeatedly to determine the order of the autoregressive model. You may start with a large value of P (say 4 or 5, depending on length of series), and then keep fitting simpler models until the test for the last term is significant. Fitted Pth Order Forecasting Equation Start by fitting the Pth order autoregressive model on a dataset cotaining n observations. Then, we can get forecasts into the future as follows:  In the third equation (the general case), the  EMBED Equation.COEE2  values are the observed values when  EMBED Equation.COEE2 , and are forecasted values when  EMBED Equation.COEE2 . Choosing a Forecasting Model (Sec. 13-6) To determine which of a set of potential forecasting models to use, first obtain the prediction errors (residuals). Two widely used measures are the mean absolute deviation (MAD) and mean square error (MSE).  EMBED Equation.COEE2  Plots of residuals versus time should be a random cloud, centered at 0. Linear, cyclical, and seasonal patterns may also appear, leading to fitting models accounting for them. See prototype plots on Page 695 of text book. Time Series Forecasting of Quarterly or Monthly Series (Sec. 13-7) Consider the exponential trend model, with quarterly and monthly dummy variables, depending on the data in a series. Exponential Model with Quarterly Data Create three dummy variables: Q1, Q2, and Q3 where Qj=1 if the observation occured in quarter j, 0 otherwise. Let Xi be the coded quarterly value (if the first quarter of the data is to be labeled as 1, then Xi=i). The population model and its tranformed linear model are:  EMBED Equation.COEE2  Fit the multiple regression model with the base 10 logarithm of Yi as the response and Xi (which is usually i), Q1, Q2, and Q3 as the independent variables. The fitted model is:  Exponential Model with Monthly Data A similar approach is taken when data are monthly. We create 11 dummy variables: M1...,M11 that take on 1 if the month of the observation is the month corresponding to the dummy variable. That is, if the observation is for January, M1=1 and all other Mj terms are 0. The population model and its tranformed linear model are:  EMBED Equation.COEE2  The fitted model is:   EMBED Equation.COEE2   EMBED Equation.COEE2   EMBED Equation.COEE2   EMBED Equation.COEE2   EMBED Equation.COEE2   EMBED Equation.COEE2   EMBED Equation.COEE2   EMBED Equation.COEE2   EMBED Equation.COEE2   EMBED Equation.COEE2   EMBED Equation.COEE2   EMBED Equation.COEE2   EMBED Equation.COEE2   EMBED Equation.COEE2   EMBED Equation.COEE2   EMBED Equation.COEE2   EMBED Equation.COEE2   EMBED Equation.COEE2   EMBED Equation.COEE2   EMBED Equation.COEE2   EMBED Equation.COEE2   EMBED Equation.COEE2   EMBED Equation.COEE2   EMBED Equation.COEE2  OPwx !#$ABYZ[\ȷ{qhj?CJEHUjB UVmHjWCJEHUjB UVmH 6CJH*6CJjlCJEHUjήB UVmHCJjCJEHUjB UVmH jCJU5CJ5CJOJQJ5CJ5CJOJQJ5CJ OJQJ5CJ$OJQJ5CJ$OJQJ jUmH( 2>?vwxAU & F$ 2>?vwxAUde ~ R ǿ~vn`  4      Xde12d      X     4 5fghij)./01<=CDHIUVmnop*4N\e5CJjCJEHUjARB UVmHjCJEHUj'RB UVmHj CJEHUjڻB UVmHj CJEHUjQB UVmHjCJEHUjQB UVmH 6CJH*6CJCJ jCJU5de ~ R  I  & F & FGIkl   P Q   ) * + , 0 6 7 N O P Q ¹ޯjo5CJEHUjB UVmHj5CJUjsCJEHUj֑B UVmHjCJEHUj\B UVmH 6CJH* 5>*CJ>*6CJCJjCJEHUjB UVmH5CJ jCJU5Q R S T ] ^      H I ` a b c f      . / w x } ~  Ǿ5CJj$CJEHUjQB UVmHj!CJEHUjB UVmHjCJEHUjOB UVmHjCJEHUjB UVmH jCJU 5>*CJ>* 6CJH*6CJCJ9  I  d e f   . /  &BCDEcd567_`a3t8y!"#$%:         #  TI  d e f   . /  &BCD & F & F & F & F  !"#$%&'>?@ABDEcd56CD[\]^aܾܻ☏⢅{jB UVmH 6CJH*5CJjX0CJEHUjB UVmH6CJ>*j,CJEHUjB UVmHCJ jCJUja*CJEHUjB UVmH5CJ 5>*CJCJ jCJUj'CJEHUj~QB UVmH/DEcd567_`a3t8y & F @ & F & F345RScd{|}~ 췮줛j9CJEHUjB UVmHj7CJEHUjˣB UVmHj4CJEHUjB UVmHCJH*5CJ>*CJ>* 6CJH*6CJCJ jCJUj2CJEHU9y!"#$%:B & F  & F & F & FBShijk)*+,-./0123`abijcdtx123kprv"$&fB D | ~ $!!!!"    \Shijk)*+,-./0123` & F  & F    :;de~ *+,-ghjkp23`bjdtxy⹰֭⧣♐jCCJEHUjʚB UVmH5CJ 5>*CJCJj@CJEHUjꢔB UVmHj_>CJEHUjB UVmH5CJ6CJCJH*CJ jCJUj<CJEHUj>B UVmH4`abijcdtx123k&d&d13"&7_jlnprtިޞ56CJH*OJQJ56CJOJQJ5CJOJQJ 6CJH*6jCJUmH6CJ 56CJjICJEHUjB UVmH5CJCJ jCJUjECJEHUjB UVmH8kprv"$&fB D | ~ $!!!! & Fhjln& .0^`bd  4 6 8 : < D F t v x z ~ ȿ揄zqjuUCJEHUjʖB UVmHjR5CJEHUjwȖB 5UVmHj5CJU5CJOJQJjPCJEHUj'ɖB UVmHjNCJEHUjwȖB UVmH jCJU 56CJH* 56CJ5CJ5CJCJCJH*OJQJ CJOJQJ,~ !! !"!H!J!x!z!|!~!!!!!!!!!!!""""""O"s$$$$$$$%%%%%%=%>%Ǿڭ룚|j5CJUmH6H*6jCJUmH5CJjz`CJEHUjE̖B UVmHj^CJEHU5CJOJQJj[CJEHUj'ɖB UVmHjGYCJEHUjq˖B UVmH jCJUCJ 56CJH* 56CJ5CJ0!""M"N"O"n"""##$$p$q$r$s$$$$$$$$%%%%% & F""M"N"O"n"""##$$p$q$r$s$$$$$$$$%%%%%%M%N%P%Q%%%%%%%&&&&\&]&^&_&a&b&c&d&e&f&g&h&j&&&&&&&&(()H)J)**+++(+)+++++++9,:,<,r,s,/-0-1-Y-Z-[------.n.. b%%M%N%P%Q%%%%%%%&&&&\&]&^&_&a&b&c&d&e&f&g&h&j&&>%J%K%M%N%O%%%%%%%%%%&&&&@&A&X&Y&Z&[&_&`&j&&&&&&&&&E'F'o'q''''''''''''''''P(R())) ) )) CJOJQJCJH* 6CJH* 6CJH*6CJCJ 56CJ5CJjVd5CJEHUjNB UVmHj5CJUj5CJUmH 56CJH* 56CJ5CJ>&&&&&&&(()H)J)**+++(+)+++++++9,:,<,r,$))))))0)2)4)6):)@)B)D)V)X)Z)))))))**z*{********++(+)+D+E+k+l++++++++9,:,;,<,s,1-=->-U-V-W-X-[----jTfCJEHUjB UVmH jCJU5CJ 56CJH*jCJUmH 56CJ5CJj5CJUmHCJH*6CJCJH*OJQJ CJOJQJCJ@r,s,/-0-1-Y-Z-[------.n...../8/o/p/q/r//0 & F & F & F @ & F--------.........///{/|/}///<0=0N0O0P00000,1/101E1F1d1e1|1}1~111111111111111U2g22ǾjqCJEHUj6B UVmHjnCJEHUjB UVmHj5kCJEHUjB UVmH5CJ5CJ6CJj9iCJEHUjB UVmH jCJUCJ 6CJH*<..../8/o/p/q/r//0<0=0>0?0@0U0000000000000.1/101E1F111111V22 33333344f5h5555506162636465666768696:6;66666688d8f8888869899999:::R:T:V:X::x;<<<<<d0<0=0>0?0@0U0000000000000.1/101E1F11111 & F & F & F11V22 33333344f5h5555506162636465666768696:6&d22222233 35373I3L33333344444,4.4p4v444444^5b5p5r55555555555 6 6+6,6/606;66666666888 5CJH*5CJOJQJ5CJjAvCJEHUj)B UVmHjsCJEHUjB UVmH jCJUCJH*OJQJ CJOJQJCJH*5CJCJ6CJ 6CJH*>:6;66666688d8f8888869899999:::R:T:V:X:: & F88 8 8888888 8h8j8n8p8r8t8v8888888888888889.9092949v9x9999999::::::J:L:żj֚B UVmH 5CJH*5CJOJQJ6CJj9}CJEHUj%B UVmHjzCJEHUjB UVmHjxCJEHUj^B UVmH jCJUCJH* CJOJQJCJH*CJ5CJ4L:N:P::::J;L;N;P;;;;<<Q<R<}<<<<<>>>>? ?@@@@@@@@AAAAAAAAAAAABB/B0B3B?BBBBBBBBBB CǾ촫եե 6CJH*jOCJEHUj'B UVmHjCJEHUj5B UVmHCJH*6CJ5CJ5CJCJH* CJOJQJCJ jCJUjCJEHU?:x;<<<<<<< < < < < <<P<Q<R<}<~<<<<====> & F & F & F<<< < < < < <<P<Q<R<}<~<<<<====>X>>>??@@zA|AAAAB4B5B6B7B8B9B:B;BB?BBBBBCCC,C-C.C/CKCLCMCCCCCCCDDDDDE@EFFFFFFFFF*G,G.GHHHHIIIIII/J0JJJd>X>>>??@@zA|AAAAB4B5B6B7B8B9B:B;BB?BBBB & FBBCCC,C-C.C/CKCLCMCCCCCCCDDDDDE@EFFFF C C C CCCC(C)C*C+C/C0CGCHCICJCMCCCCCCCCDDDDDDEE4E6E8E:E~EEEEEEEEEEEEEE⽴⦝Ⓤ CJOJQJCJH*jlCJEHUjެB UVmHjՓCJEHUj=B UVmH6CJjCJEHUjLB UVmH5CJjCJEHUj֨B UVmH5CJCJ jCJUjЈCJEHUjB UVmH2EEEEEEEEEZF\F`FbFdFtFvFzF|F~FFFFFFF,GGGH HHHHI I!IIIIIIIIIIIIIII J JJJJ+JƽjCJEHUjY7B UVmHjCJUmH5CJ6CJ5CJjŚCJEHUjTB UVmHCJH*OJQJjCJEHUjҭB UVmH jCJUCJCJH* CJOJQJ9FFFFFF*G,G.GHHHHIIIIII/J0JJJJJJ.K/K1K$+J,J-J.JXJYJJJJJJJJJJJJ.K/K0K2K3K5K9K:KQKRKSKTKVKWKYK]K^KuKvKwKxKyK{KKK$LjLMM`NaNbNcNdNNNNNN⮥jOB UVmHjCJUmH5CJjHCJEHUjJ:B UVmHj3CJEHUjH6B UVmHj5CJUmH5CJCJH*6CJCJ jCJUjrCJEHUj9B UVmH7JJJJ.K/K1K2KUKVKyKzK{KL L!L"L#L$LiLjLMMMM`NbNcNdNNNNNNNNOOOOOO"P#P%P`PaP}P~PPPPSSSSSS%S&S(S)S*S+SUUUUUUUUUUUUUUUUUVVVV4W5WWWXXZZC[D[[[[[\\d1K2KUKVKyKzK{KL L!L"L#L$LiLjLMMMM`NbNcNdNNNNNNNNNNNOOOOOOKOLOQOROTOXOYOZOO"P#P$PFPGPaPbPyPzP{P|P~PPFQQQQQ$R2RRRS&S'S1S2SISJSKSLSSSSSSSST TUU쾵죚Ք 6CJH*jCJEHUjVB UVmH5CJ5CJjCJEHUj[QB UVmHjCJUmHCJH*6CJjΪCJEHUjNB UVmHCJ jCJUj1CJEHU:NOOOOOO"P#P%P`PaP}P~PPPPSSSSSS%S&S(S)S*S+SUUUUUUUUUUUUUUUUUUVVVV4W5WWWXXZZ & F$UUUtVuVVVVVV4WoWrWWWWWWW.X/XzX}XXXYYbZeZZZ>[A[[[[[[[[[ \\9\=\W\a\\\\\\\\\\\*]+]@]A]X]Y]Z][]f]g]~]jCJEHUjYB UVmH 56CJjCJEHUjB UVmH jCJUCJH* 5CJH*5CJ6CJ5CJCJCZC[D[[[[[\\]]\]]]]]]]]]]]^^p_q____ & F & F\]]\]]]]]]]]]]]^^p_q____``1`2``````3a4aEaFa7b8b:bbbbbbbbbb2c3c5ccccccccccddddee>e?eUeVeeeeeeef fCfDf*g+gPgQgggggggg i iKiLiiiiiijj0j1jd~]]]]]]]]]]]]]]^q________`2`3`J`K`L`M`N`````3a4aEaFa+b,b8b9bbbb2c3c4cccccȵȠjaB UVmHjCJUmHj˼5EHUjB CJUVmHj5EHUjB CJUVmH j5U55CJ5CJ 6CJH*6CJ 56CJCJ jCJUjCJEHUj_B UVmH4_``1`2``````3a4aEaFa7b8b:bbbbbbbbbb2c3c5cccccccccccddddee>e?eUeVeeeeeeef fCfDf*gcccccedfdgdddddddddddddeeee"e#e:e;e~t#& .މaӫLM΅v|~|OrP}`!hAd!6 ^_p Ƚ ixcdd``c!0 ĜL0##4Y=F (.2KA?H 6XnĒʂT.(X'#M`Ay`!0QrR#b%`( "x5Rx_hG;owws˟і4<h)EД<L±<DJ-(r zXćB)Ml|Zl3 uawg?74@.d?`.FQJj[SAf1[k>SHC n' ݴ5]{GYC;M`Wо(:$+%ҝ`]]bD5\vh\1u&)ؘeV1tl#Y|^Wg_xd@<7l/dvKS7 %Δih`>QN(_MwwOo J0cnMwj@ќ~5tLQFzFV566ddUߴx7*ͩy|_`nd*Q&mtzY#ԛ=ڴZzd?d_ʤ+;!|rO+d(%ThSX5 66ɪ><6qUC6R} 3(tGn"e5[N@LGN09bbdls14}:鐻A{䏰 sj{3JJ}V%͸/A} R޵P}E6_'dI\ U RyҘM>ߊ S |vi eΦ3!u˔4&M2鿆de.,!WLRt%b|. :2ʿ3 $] dҡudZB:L^Dw >u{lb7øgU cJ,t>pmXK$ӞpQOP !nkx?Ͼth^%k7 8 !L`!R/ɺnRi*L`x :xSJAKr%`0"ˆR@(0p,* ;_BDL <粁D$ƅٛnfv-leAqzΒq *,YG@avǭq`%WNj YH1bF^yAO7'ֿE̩Ra\-Xz+]Ӳvs HY2 <2Ommwu&n׾_IɸygJʬ5ev+o4/yW9!ı.$lsQBѐ8G֭اU-ݫ!M k,d|0ϳsP5⋮}k&>ǚ0{f?Xsz͓;h4ݦB߸zђU &ۭz# WczVo':c%`!%Z,Dw?.!h(3 xcdd`` @c112BYLGcbAէNW]4lw-M9e}wx/ ޳a/\/ ޸J׌&`!8AR?6KC:&C< xKQܙuG֟kilCeKK!.Aઠ IRP}ؠ*QzHPRv{ðϽ{ VHU;R\.'RuX (hlxyG\4ƞSGNX#q(S6yW^7q&ca,va?nRktug 'É!x6Ň?$ PpfXzD"ݐeREʟ1f%-Qm4ƴHѼ oe9niZ܉JW9qq:l2.#\ՖWղijUu✴~^9}DA|8");vhp"Vxa ns~bSֈ9'"B8D>Xm3ĉ&9DI#k*Fp{X,Kzv2 Es缪59{#k*FHNͽ/2l{SVzLDҋadb 3q{kFqȍJ|"9yw7^Btw~ytC,[/%d&w)# ]ɸT$i}>C*'O׈$eI/6DS暝kޗ&}{E2j b̄B'ӝ?Mʕ֨'em`!FY_ʵ {36&@<}xMhA̤ UXӺjn؋'% z!VZ =kOE'/{XEAԓ(([!hMR4aٙo޼ `G_T(!^v'qYfnjRawyR->bӄX~\Ι)<AӢ o˼Lx\T5*fv/Ã']!e\ J3>.3}xyKAMwd椷Í!ų(1,Ȕ$65L2&I9OT] 5nRaKN4 Rvf8eGk)]OGk1]۫H8\t Nijr-ֻ\r)*}S;fuأuسA]d5e܍b6EKKW eČbr"I# >_r,˻SwX 2ែ& !8֌z5$ݥ9=5vEfk&{ېw}doY߄Rvf %3 J,vo}<|\`H8`!͔KVTd .Xxtxcdd`` @c112BYLGcbc0r``錬 g$UaEcDTYSqh?ݜΤ6!*i/c{+"A`"C"Fm/ -@&W&0;sr#ίLL #l3/EDD:9Ex`"aYY1D䀉BZh(V%2@B8Q2' ?y-m `p>adbR ,.I ca, `!W9b#}E%*HA%xQHSQǿsyńfWk!|01ȗ""[:*BRo=[C1!F"HDJd^Ru=^ώq}瞳@r/vѳ[Fȴf&TnZE6s14xwCdxN{C{b9##$'˘{Fcp;R̬Rj=ZlbPƪX%>O$NCꟘJ@ Y&< T_>26~?S+"D*Z*5SdZ3v8ҺS\ 2-BtcnpE&02D$fK I[2yfM4b8Et2G&M1K[J*]u29_oiԱyj\3OB:LEM7\BM\) aLs5Fށ\'UOԂ'L+7v\qnk,nfAr}YGV8*xS =P`\=2Ƶa2<O+jࢹuq51leѡ,FGϲ]{W`K-vjNZt`A;u[d/|owWލ^rq"N&2TT4ɷ#ٹRY18w5Y53"= X`! 43PGQ?M( x=K1ߤ_kCu(IqMRV G6gAppupuSn I.A {${a  hc79–eI#(Di@0=`@cFwӘ8k0%:g{cuMhn-[(n$˛PT yc. Q ؛9R$bd< D`d#i-g]Ȥ L^Zdr$w[SǐQ )lm1 uJنHq R32!Ga[erq 5fRWJb-dn~ce$r1Z;qrߒpOl9}ɌGDi&sߒ`}L m7Jkl^4>XVɘaELX'}掟G5GȚd@ٮ ,sCUH3L'i>uw;Mqƨ u2N{E{_/ʮ^w2xhy+ v}&?›FI' w`!'C4T3}Ǟ @H|xcdd``gb``baV d,F1J@Yp1&n! KA?H _rC0&dT20 rosv Й! ~ Ay ۏ/? N300}`ĵQ3*\;H$rd?b`+*A $QƁ* ➄dF: &EpD`j&]I9 @4w}^km+/C 9/ Y@FhWVEa"0*! aEy<}z8.`)h CZ DްāӪH)'L y@=|Ĥ\Y\ q1(2tA ;/Gf,H\`!PAShlkX@`\x=K@wI[ꢢPvr)DA-(8Ҡbk b'p|! "Ap((ƻˋ\TL_Ce PC\4(!DqK!KPGϧ4M0{Kgbun$0^*Uh~ yHfd^0bӛ[+Ykqe^F!,W VtՊ}Ĵꟲ' hy|Rh2 jNyH؍C=G4K&v;М ͌]p !a龷݌9 lݻg^{ɀ;EE%Q)sPv]֛B2J0)ݐH6"ir꿝ïgِ31R5thޫe30C,/8Z4io`!"4"}2`0, PxRJA};0(hJ*&`c9B,l,ElXٽM `޾yofnL80Hy؝# “i#CİThlZ7F~| Rq(h2]Ss!ߪlN.W+m\߷&G`3Z]p @r)sV Tq'w#D,_M$Jμf^lt^ʉrNeZa´r?u{VviՐscdR'U5Y: u Me5S/Q[u_DΩ70}#קnZv3Cyi]`! MOIJ@)(?RxMhA̶iIШMb?M?~T)(YzB[K+-[he z؋X=$7-[U"<)^Du&y._?mA@ :f1"\E1?bu]hb" A\AB<}bωAqvde‚&1罍bWC%EU1;\mXOn+}nJCbr\Z4,c|<{Y1 _.#OoKB)s# {1EәNivQrǠdoxu| ̣>qX %/tJI9J6()R1K"Ş{s9ZL; o5$:9):`:>YFb6:XPtkuў:=ͅRұHyMsSr8TOk{J9 뫛X3Z1׃obwm$6MQU.Pր"QzX^ܖab=ǽjFP?U=RO ";%@"FH"7%A"H"[eA"H"`\B& +ss\ąO9pZNtX^6]~.H83r $0X,2'!iRj c3|1&M$zp D&Eڥ`"0|@okEIQ"ħFĦ`#Aed^Bl``K(F&&\A ~<fyx9 h`!l6LX?czZA FCe\6/MNvmL38IR ?Z{_ ,'c?'xnfadI|Ut=ǕA2JVxGy+vQX(I%)vUg̜fX q v,Lt12'ދhDaO_å*eOdB6^&ܳˤهffB:8w1eG&/\Ns5\lѮZakyGQy>y6d dcp1ŀc7Լf5|_UQ[nʶ޺dH"2I>IC^\ YGA$ ߽yXI{r$D5s52ɳN*G2^QLLv)>V&I_} >s3k0C×gӤ4NS똦+δteo1^wRKEO=G ;L>Sv2/ϏqJgJ=g`!{G=P ' xkA&Gj!*HִzSbO(jR &H z("^ԓ "*C$To/"Eu~&sښOoE(mO:#;Bq,tGa8SdM&phu1XH1TlY+B 4_x=-Nr4>RjW+Siȑn-6,N69g(`zA|u)tG(N_;Zڈ(>'ePgAg1-{Uq[*I@JQJr]5J.rs@*$)"_EzZJ{_h[ܳr?|4ī#]|7x>"3쉊(3FDwXNXo<a1m=d|xT֌(9f A˚+ YswYNjlmD77ll}ɚeQ/YQx5TSc=25IW# Ys0,kޚADh]nh5^{a:sY@#O<~^k# ̟])}2Z3n_kFnunw|KYd3dr>`39AcW; aМ*ʀpϏ xӐ}UXfJlDdD@B  S A? 2[j#!_V `![j#!_VH  xxcdd``^ @c112BYLGcbYA&dV1pAc B MN@o*Wx\ ]` ca֎!Dd@B  S A? 2kn~ K`G`!?n~ K`H  xcdd`` @c112BYLGcb?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~?Root Entry: F- >Data sCWordDocument9\ObjectPool<σ> >_1116581122Faσ>3Ѓ>Ole CompObjjObjInfo "',1458;>ADGHKNORUVY\]`cdehknqtwxy| FMicrosoft Equation 2.0 DS EquationEquation.COEE29qaD Y i =b 0 +b 1 X i +e i FMicrosoft Equation 2.0 DS EqEquation Native }_1116581582 FЃ>Ѓ>Ole CompObj juationEquation.COEE29qD Y i FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD X iObjInfo Equation Native  4_1116582292FP><у>у>Ole  CompObj jObjInfoEquation Native 4_1116584344 FJN҃>JN҃>Ole CompObjjObjInfoEquation Native W FMicrosoft Equation 2.0 DS EquationEquation.COEE29q;D b 0 +b 1 X i FMicrosoft Equation 2.0 DS Eq_1116688768"FӃ>Ӄ>Ole CompObjjObjInfouationEquation.COEE29qD b 0 FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD b 1Equation Native 4_1116688844FӃ>Ӄ>Ole CompObj jObjInfo!Equation Native 4_1116584922$F0"uԃ>0"uԃ>Ole   FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD e i FMicrosoft Equation 2.0 DS EquationEquation.COEE29qCompObj#%!jObjInfo&#Equation Native $4_1116688935@)FeՃ>eՃ>Ole %CompObj(*&jObjInfo+(Equation Native )4D b 0 FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD b 1_1116688961.Fp$Yփ>p$Yփ>Ole *CompObj-/+jObjInfo0-Equation Native .4_11170944223FN׃>N׃>Ole /CompObj240j FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD Y i =b 0 +b 1 X i +b 2 X i2 +e i FMicrosoft Equation 2.0 DS EqObjInfo52Equation Native 3_1116691695,;8FA؃>A؃>Ole 6CompObj797jObjInfo:9Equation Native :y_1116853785=F09ك>09ك>uationEquation.COEE29q]D 'Y ^  i =b 0 +b 1 X i FMicrosoft Equation 2.0 DS EquationEquation.COEE29qOle <CompObj<>=jObjInfo??Equation Native @C'D 'Y ^  i FMicrosoft Equation 2.0 DS EquationEquation.COEE29qrD SSE=(Y ii=1n  -'Y i^ )_11168583206cBF4ڃ>4ڃ>Ole BCompObjACCjObjInfoDEEquation Native F_1116858776GFPN'ۃ>PN'ۃ>Ole ICompObjFHJj 2 FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD SST=(Y ii=1n  -Y) 2 df T =n-1ObjInfoILEquation Native M_1116858959EOLF܃>܃>Ole P FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD SSR=('Y i^i=1n  -Y) 2 df R =1CompObjKMQjObjInfoNSEquation Native T_1116859338QFP ݃>P ݃>Ole WCompObjPRXjObjInfoSZEquation Native [ FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD SSE=(Y i -'Y i^i=1n  ) 2 df E =n-2 FMicrosoft Equation 2.0 DS Eq_1117082024|VF0݃>0݃>Ole ^CompObjUW_jObjInfoXauationEquation.COEE29qD r 2 =explainedvariationtotalvariation=SSRSST0r 2 1 FMicrosoft Equation 2.0 DS EqEquation Native b_1117081982[Fރ>ރ>Ole fCompObjZ\gjuationEquation.COEE29qdD r=sign(b 1 ) r 2 -1r1 FMicrosoft Equation 2.0 DS EquationEquation.COEE29qObjInfo]iEquation Native j_1117036983'`FP߃>.>Ole lCompObj_amjObjInfoboEquation Native pl_1116860550JreFn7>n7>-      !"#$&%'()*+.,/0132456879;:<=>xw@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvzy|{}~PD V(Y i )=V(e i )=s 2 FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD S YX = SSEn-2  = Y Ole rCompObjdfsjObjInfoguEquation Native vi -'Y i^ () 2i=1n  n-2  FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD e i =Y i -'Y i^ =Y i -(_1116862091mjFh>h>Ole zCompObjik{jObjInfol}Equation Native ~_1116862375oFp>p>Ole CompObjnpjb 0 +b 1 X i ) FMicrosoft Equation 2.0 DS EquationEquation.COEE29q%D 'Y i^ObjInfoqEquation Native A_1116863444htF`l>`l>Ole CompObjsujObjInfovEquation Native _1117037515yF~>~> FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD D=(e i -e i-1 ) 2i=2n  e i2i=1n  FMicrosoft Equation 2.0 DS EqOle CompObjxzjObjInfo{Equation Native DuationEquation.COEE29q(D d L (k,n) FMicrosoft Equation 2.0 DS EquationEquation.COEE29q_1117037556Y~F>>Ole CompObj}jObjInfoEquation Native D_1116864062F>>Ole CompObjj(D d U (k,n) FMicrosoft Equation 2.0 DS EquationEquation.COEE29q.D Dd u (k,n)ObjInfoEquation Native J_1117037213F9>9>Ole  FMicrosoft Equation 2.0 DS EquationEquation.COEE29q.D Dd L (k,n) FMicrosoft Equation 2.0 DS EqCompObjjObjInfoEquation Native J_1117037290wFk>k>Ole CompObjjObjInfoEquation Native luationEquation.COEE29qPD d L (k,n)Dd U (k,n) FMicrosoft Equation 2.0 DS EquationEquation.COEE29q_11171007461FA>A>Ole CompObjjObjInfoEquation Native 4_1117100928F > >Ole CompObjjD b 1 FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD H 0 :b 1 =b 10 H A :b 1 b 10 TS:t obsObjInfoEquation Native _1117101570TF`i>`i>Ole  =b 1 -b 10 S b 1  S b 1  =S YX  (X i -X) 2i=1n   RR:|t obs |t a2,n-2 FMicrosoft Equation 2.0 DS EqCompObjjObjInfoEquation Native _1117177975F@?>@?>uationEquation.COEE29qD H 0 :b 1 =b 10 H A :b 1 >b 10 TS:t obs =b 1 -b 10 S b 1  S b 1  =S YX  (X i -X) 2i=1n   RR:t obs t a2,n-2 FMicrosoft Equation 2.0 DS EquationEquation.COEE29qcD E[Y|X=X i ]=b 0 +b 1 X iOle CompObjjObjInfoEquation Native _1117178151F>>Ole CompObjjObjInfo FMicrosoft Equation 2.0 DS EquationEquation.COEE29q]D 'Y ^  i =b 0 +b 1 X i FMicrosoft Equation 2.0 DS EqEquation Native y_1117178374F@q>@q>Ole CompObjjuationEquation.COEE29q/D 'Y ^  i t a2,n-2 S YX  h i h i =1n+(X i -X) 2 (X i -X) 2i=1n ObjInfoEquation Native K_1117178737FI>I>Ole  FMicrosoft Equation 2.0 DS EquationEquation.COEE29qaD Y i =b 0 +b 1 X i +e i FMicrosoft Equation 2.0 DS EqCompObjjObjInfoEquation Native }_1117178949Fv>v>Ole CompObjjObjInfoEquation Native QuationEquation.COEE29q5D 'Y ^  i t a2,n-2 S YX  1+h i h i =1n+(X i -X) 2 (X i -X) 2i=1n _1117186638FУ>У>Ole CompObjjObjInfo FMicrosoft Equation 2.0 DS EquationEquation.COEE29q%D S b 1  FMicrosoft Equation 2.0 DS EqEquation Native A_1117192465F@z\>@z\>Ole CompObjjuationEquation.COEE29q D e i =Y i -'Y i^ =Y i -(b 0 +b 1 X 1i +L+b k X ki ) FMicrosoft Equation 2.0 DS EqObjInfoEquation Native _1117193459F$>$>Ole CompObjjObjInfoEquation Native _1117193405F->->  !&)*-0367:=>?@CFILMNORUVWXYZ]`abcdefilmpstw|uationEquation.COEE29qD SST=(Y i -Y) 2i=1n  df T =n-1 FMicrosoft Equation 2.0 DS EquationEquation.COEE29qOle CompObjjObjInfoEquation Native  D SSR=('Y ^  i -Y) 2i=1n  df R =k FMicrosoft Equation 2.0 DS EquationEquation.COEE29q_1117193526^FpM>pM>Ole  CompObj jObjInfoEquation Native _1117194233F@>@>Ole CompObjjD SSE=(Y i -'Y i^ ) 2i=1n  df E =n-k-1 FMicrosoft Equation 2.0 DS EquationEquation.COEE29q=D F obs =ObjInfoEquation Native Y_1117194281FR.>R.>Ole MSRMSE FMicrosoft Equation 2.0 DS EquationEquation.COEE29qID F obs F a,k,n-k-1CompObjjObjInfoEquation Native e_1117296734FR>R>Ole CompObj jObjInfo"Equation Native #( FMicrosoft Equation 2.0 DS EquationEquation.COEE29q D  FMicrosoft Equation 2.0 DS EquationEquation.COEE29q_1117296814F0{ >0{ >Ole $CompObj%jObjInfo'Equation Native (_1117296933F( >( >Ole +CompObj,jnD t obs =b j -b j0 S b j  FMicrosoft Equation 2.0 DS EquationEquation.COEE29q[D |t obs |ObjInfo.Equation Native /w_1117297366Fз >з >Ole 1t a2,n-k-1 FMicrosoft Equation 2.0 DS EquationEquation.COEE29qfD b j t a2,n-k-1 S b j CompObj2jObjInfo4Equation Native 5_1117468084F>> V8p( T W5I&*&d++&1`7L|Q: \Pp 1b;'F&&\A ~<fyx9KDd@B   S A ? 2l{CCw_@0H`!@{CCw_@0 xcdd`` @c112BYLGcb=^8g?&IHvcEIBV/S j00]caPY1D#K^4UsS>97 0y{I)$5bPd+? 32z`! Xni>2 @Xhxcdd``bb``baV d,F1J@Yp1&B@?6 p30fꁪaM,,He`d=k%@g d++&1l?z+8U`T T090-cve8@>`Yy. ح?,& Y 2+"_e`bUuᄩ{'27)?(0WX@3pS.PO. Cl``<#RpeqIj.ŠQ ? 3?p6#>av&䶾ii9iSV}^ٸGǟx6+Dd B  S A? 2AnW+K]es+!`!AnW+K]es+`x+aǿ3k=HA9('EbsȆX++\="J'()u\\9:82ޟks!4>?f!X (C*!ccAN֠`SE{ :Ƚ7]&{m1[2N* #zߌuX7Un,&lC"zXI &#r |O9Dk ,o|͟wZd;Y(+D杙+ΖZ@< `v0’TyFVE0;N`UzVi(}>j2v0mS MQkb[U]Xc 5cNNL)7.vzEf~xhKs/93Qm2.b4T$zVtch %2bd}>w#l)ƽz/wf_y]/ע'<qDd B  S A? 2@-ڕ#yHc`!@-ڕ#yH2xtxcdd``db``baV d,F1J@Yp1L,56~) m @ sC0&dT20 rosv Й! ~ Ay ۏ/? N300}`dҙZ . $Yy.Rˊ.. I<CDT^4YgXt J;\ft}6tO"0Xn> +ssظ _ ~ meH HDID]^Nt5̼ mz2P/c92#mf2qШl&PX$2 .r "\<w 0y{yI)$5bPd(v<^Wf,G[JDdB  S A? 2frKm_>fǠhB`!:rKm_>fǠhxtxcdd``fbd``baV d,F1J@Yp1L ,56~) m @ F.熪aM,,He`@L 53B2sSRs=^~*g`0``**ˁI3*Dd?M^t lrf $b2# =`*3I=bdbR ,.IAď ;/G3X?Rb+DdXOB  S A? 2X3-b\b;@4`!,3-b\b;@  70, x;hQ{3K\L#c1jaRawpItZ,B"D:wCԁ;s7s^?C x}7b1f,FeYޞdtEèhV'<#(b_X&0 M x |'!/_)cyē֨jd?hy=ez7z}<{bś8e3f¿ y1M6h%$d$pаDבRϮ7]~ёͨJ_EC4k^۠J\h Ճb.G~ICqrOj.ȩZp8 ?\SXVu3NU,uecOg_vOϔu:*GOxXRQll_8·'>W[UvRfvd2<H89Cum4:ٌhb%85sFv^WSDd[@B  S A? 27PG"Pd 0[qbR`!Z7PG"Pd 0[qL`S (xcdd``Vcd``baV d,F1J@Yp1&V! KA?Ha@cjx|K2B* Rl7Sÿ9`wMLbrc0J`4aL<)adWb@lUM``|*k0@ &3|lFe\ &P 41"v0o8+KRsA<.E.b0΃ 3}k(Ddp@B  S A? 2j!5WJ]Yan`!Yj!5WJ]YL 'xcdd``Vcd``baV d,F1J@Yp1&V! KA?Ha@cjx|K2B* Rl7Sÿ9`wMLbrc0J`4aL<ɮ\y ت?,[!!Z? $b2#Ϳ&kT6EЩLahpAC #!.``CI)$5bPd+? 3 ."0oL p5¬ UpL6HP  r#wK J+ >5l.ܤr'W\>wA6W^;ܜX2B.b: p}v&P 4f"v0o8+KRsA<.E.b Xa~eLDd@B  S A? 2!|JjoWx`!!|JjoWx  Xxcdd``ad``baV d,F1J@Yp1&&! KA?H@6XnĒʂT$@o].:S $37X/\!(?71aXr2]b}@:]*Ea"ߘ@")jXA"ٕ"=l <;GF0ϙA&T}j2\I9 @\.|1`-v 9q 8|t*"A M.hrcb9E `p 321)Wx\ ]@0΃ .傂sDd$ @B   S A ? 2Ě˯s.{r}`!Ě˯s.{r}H xcdd``^ @c112BYLGcbiFZLV:iQ7x$k 1*kdi3;~7~U;=Ǧ:"Tw{QH0}gU:6Hhw{'E~RU,K58,gQ A29g.4{uk\qQRNнAŢfD\%jYh _qtqV%-XMaZI~W ;Ztz7ABT_G73a-by}L&4/ajDd B   S A ?  2nd$luq `!nd$luql  0h+vx[HQǿ9sˮVL=I$+HȎD/)J(,T=Pat mADQO %H$+˜Ygla3|C @@>2>| +yB;,=W{RCA>,>2589|=x#jB,|pkP']&^u5+?kVjÌ]SAԛAp籂iUנ[bZ^OsCvλ[ GQ:w\xϐړ4b~xr.yNB~[8$%U!pωNIT>d2v\T6Q }61e.kZS%4 Q_JGmuiu\tHU5'~v-We Q!a斑?)$@)6#!S{$@Pea+gu%ɔ|6MZzVI9i¸?Dk5i!7YW(v^9u+N?81HXCN(nXtGii&=;*| P ڙ wd'MyMJJclA8>wn ~2MkfZdvH#pCzQǼtZ7 1"Tk$Vb=wYorQ/1Đf4NtӧށIN3.vIqd\ !Qyƣ11z/hP[3msks֛OVbGXVf6t,RcnZkFۈNXg@gO鿴n354'g: TL2|sId" ԞDd @B   S A ?  27 14!oQt`!7 14!oQ xcdd``~ @c112BYLGcbc0J`at|96+*\Ċ$F ]Č&qh\?]xюdBf\ă& $.F/lbTE T} v&I9 @\}4 p"A"@a& '$$D㔴pƌ5\ # * "} C4ss* `p`dbR ,.I ca, ±[DdB   S A ?  2:L{s&Td[ `!:L{s&Td[$@ xtgxcdd`` @c112BYLGcbc0J`at|96+*\Ċ$F ]Č&qh\?]xюdBf\ă& $.F/lbTE T} v&I9 @\}4 p"A"@a& '$$D㔴pƌ5\ # * "} C4ss* `p`dbR ,.I ca, ±Dd$B  S A ?  2<U^C `!U^CƜ,HD xOHQϮ誛=D+C%zIB=HUPK!0!ae+rK K/^$cn_+u7crҬHU{5yw5n<<>3pq5{?}2*r)ud=hVEPyD퍅pYmMAFsټYgU\qvs}m޽5ṍ}n$^Ag}$xTڮ=^WH϶89K8XT] kh8O:4 !f\ fz1/\(lDdD@B  S A? 2u[ 짤Tm-ۺ `!u[ 짤Tm-ۺH  xxcdd``^ @c112BYLGcbβ4e.e! KyoaDd@hB  S A? 2h  ՊBD `!<  ՊB@ | xcdd``fbd``baV d,F1J@Yp1&&&! KA?Hc@P5< %! H6&]'ZZǰ@pT9 P.P1g F\  &*&DF0nF. IP;BMf ѭ&`LlX %.pHB &#RpeqIj.ŠV "~<fyx9?`/DdB  S A? 2OEB$W[=\+ `!#EB$W[=\ `'xtxcdd``ncb``baV d,F1J@Yp1L B@?6  깡jx|K2B* Rl7Sÿ9`wMLbrc0r`2aL\y eEq@$UHdD"*_EEP4Y>d>]t J;\ft}6tO"EnήLf<ERvf2<@I9 @\fb FxmeH HDID]^Nt5̼ m$e""0`(6^0<LLɘs dFf\HD(B]_,`aY9 NH"e0O?׊DHJ ֧@Fd60rAK .p6v0o`dbR ,.IAď ;/G+3X?c;DdB  S A? 2frKm_>fǠhB `!:rKm_>fǠhxtxcdd``fbd``baV d,F1J@Yp1L ,56~) m @ F.熪aM,,He`@L 53B2sSRs=^~*g`0``**ˁI3*Dd?M^t lrf $b2# =`*3I=bdbR ,.IAď ;/G3X?Rb+Dd B  S A? 2Q3٬7b"[Zi- `!%3٬7b"[Zi@p x1HA]=wxx BLq\\-;Ie-TPX"MRkqu XL MN AHq쭺ą{{onHzpd+BDa&},3f]`n/E=ː N,rtziT[(AX ~l9a 2sܰg]׬jאH*7^>lmp.w@.vV"nD&ﰕ8^J2+Tш4ErC۬Wm0)xG~SU"(r__cƏ;Xb߉Ƀ&sd'})'iO?NfӴ~fWk;]Y{rR+k"y_iCS6deHZ.VgiM1yBe yW/aZ)=YH9P-.' `!TBIv*o>` @|xcdd``> @c112BYLGcb n"&,0 ɲLj|QO*WK8oYA&gW&EֳD* "{aj`>uM@WW&0;qs}2bsek\uanY.A/Xكk dѩ_\tN!.``/#RpeqIj.ŠQ ? 3XF2 , `!>XF2  pxR=KAݻ$3`,D,N +#AE& & ď3 `I#F,B"M@)Es#L % {7v̛%P~` ҾE .iiMX!ODk!Y$uT>1*TKY6я%|R:B)zJ; hA5/Z1j[f6clj;n{qҒ0`YeXd=H* )D4u5Đ ""JQ1ΜW8rg 2c>]oaQ-cj!JE@;y9Z8t4~j)Xsg\"\A#_ r#L Ɯ`V=E"["_+t>VN:9wf$<>>Dd<B  S A? 2rG*m\K. `!rG*m\Kd`` :xS;KAr B,T4(XZZM>b$?(NPvKQݻ) s;|()d>"yHm۷;KF]/*f:̼$agkyt@WX{&kNV N6eSf3=I`^W ?D)B<[2jaR W]b[&[%SCtU"eCāb~\t[d(k>*(U S-%PZU¦Tw)ԕiU"w!iA$Ҵ샹̀=:R]U;GڐJC\gTX#z''9B>Ole ACompObjBjObjInfoDEquation Native Ee_1117300730 F [> [>Ole GCompObjHj FMicrosoft Equation 2.0 DS EquationEquation.COEE29q,D r Y212 =SSR(X 1 ,X 2 )-SSR(X 1 )SST-SSR(X 1 )=SSR(X 2 |X 1 )SST-SSR(X 1 )ObjInfoJEquation Native K6_1117300950F > J>Ole P FMicrosoft Equation 2.0 DS EquationEquation.COEE29q|D r Y3122 =SSR(X 1 ,X 2 ,X 3 )-SSR(X 1 ,X 2 )SST-SSR(X 1 ,X 2CompObjQjObjInfoSEquation Native T_1117301068F`Ѓ>`Ѓ>=#$ )=SSR(X 3 |X 1 ,X 2 )SST-SSR(X 1 ,X 2 ) FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD r Yk12.Ole [CompObj\jObjInfo^Equation Native _..k-12 =SSR(X 1 ,KX k )-SSR(X 1 ,K,X k-1 )SST-SSR(X 1 ,K,X k-1 )=SSR(X k |X 1 ,K,X k-1 )SST-SSR(X 1 ,K,X k-1 )_1117301821 Fl>l>Ole gCompObj  hjObjInfo j FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD Y i =b 0 +b 1 X 1i +b 2 X 1i2 +e i FMicrosoft Equation 2.0 DS EqEquation Native k_1117301982F>>Ole nCompObjojuationEquation.COEE29qD 'Y ^  i =b 0 +b 1 X 1i +b 2 X 1i2 FMicrosoft Equation 2.0 DS EquationEquation.COEE29qObjInfoqEquation Native r_1117302226F`H>`H>Ole uCompObjvjObjInfoxEquation Native y(_1117302356&F>> D  FMicrosoft Equation 2.0 DS EquationEquation.COEE29q D Ole zCompObj{jObjInfo}Equation Native ~(_1117468505IF@!>@!>Ole CompObjjObjInfo  FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD E(Y i )=b 0 +b 1 X 1i +b 2 (1)=(b 0 +b 2 )+b 1 X 1iEquation Native _1117469182#Fpo#>po#>Ole CompObj"$j FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD E(Y i )=b 0 +b 1 X 1i +b 2 (0)=b 0 +b 1 X 1iObjInfo%Equation Native _1117468232(F@%>@%>Ole CompObj')jObjInfo*Equation Native ,_1117469258!5-FЁ@'>Ё@'> FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD E(Y i )=b 0 +b 1 X 1i +b 2 (1)+b 3 X 1i (1)=(b 0 +b 2 )+(b 1 +b 3 )X 1iOle CompObj,.jObjInfo/Equation Native  FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD E(Y i )=b 0 +b 1 X 1i +b 2 (0)+b 3 X 1i (0)=b 0 +b 1 X 1i_1117474756+?2Fpd)>pd)>Ole CompObj13jObjInfo4 FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD Y i =b 0 X 1ib 1 X 2ib 2 e iEquation Native _11174744777FZ*>Z*>Ole CompObj68j FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD log 10 (Y i )=log 10 (b 0 X 1ib 1 X 2ib 2 e i )=log 10 (b 0 )+log 10 (X 1iObjInfo9Equation Native _1117475163 <F+>+>Ole b 1 )+log 10 (X 2ib 2 )+log 10 (e i )=log 10 (b 0 )+b 1 log 10 (X 1i )+b 2 log 10 (X 2i )+log 10 (e i ) FMicrosoft Equation 2.0 DS EqCompObj;=jObjInfo>Equation Native _1117476600:DAFp,>;->uationEquation.COEE29qD ln(Y i )=b 0 +b 1 X 1i +b 2 X 2i +ln(e i ) FMicrosoft Equation 2.0 DS EquationEquation.COEE29qOle CompObj@BjObjInfoCEquation Native 8D r j2 FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD r T2_1117555203FF2.>2.>Ole CompObjEGjObjInfoHEquation Native 8_11175552890qKFPS10>PS10>Ole CompObjJLj FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD r k2 FMicrosoft Equation 2.0 DS EquationEquation.COEE29qObjInfoMEquation Native 8_1117555807PF)1>)1>Ole CompObjOQjObjInfoREquation Native _1117556884NUF J 3> J 3>D C P =(1-r k2 )(n-T)1-r T2 -(n-2(k+1)) FMicrosoft Equation 2.0 DS EquationEquation.COEE29qKD Y i =T Ole CompObjTVjObjInfoWEquation Native gi C i I i FMicrosoft Equation 2.0 DS EquationEquation.COEE29q[D Y i =T i C i S i I i_1117557247ZF4>4>Ole CompObjY[jObjInfo\Equation Native w_1117610474Sg_FP$6>P$6>Ole CompObj^`j FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD 'Y ^  i+1 =E i =WY i +(1-W)'Y i^ FMicrosoft Equation 2.0 DS EqObjInfoaEquation Native _1117610994dFŘ7>Ř7>Ole CompObjcejObjInfofEquation Native }_1117611087bliF 9> 9>uationEquation.COEE29qaD Y i =b 0 +b 1 X i +e i FMicrosoft Equation 2.0 DS EquationEquation.COEE29qOle CompObjhjjObjInfokEquation Native y]D 'Y ^  i =b 0 +b 1 X i FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD Y i =b 0 +b 1 X i +b 2 X_1117611190nF 1:> 1:>Ole CompObjmojObjInfopEquation Native _1117611220]sF c;> c;>Ole CompObjrtj i2 +e i FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD 'Y ^  i =b 0 +b 1 X i +b 2 X i2ObjInfouEquation Native _1117611332xFN=>N=>Ole    #&'*-.147:;>ABEJOTWXYZ[\]`cdefghijknqrstuvwxyz} FMicrosoft Equation 2.0 DS EquationEquation.COEE29q1D Y i =b 0 b 1X i e i log 10 (Y i )=log 10 b 0 +X i log 10CompObjwyjObjInfozEquation Native M_1117611603v}Fv?>v?> (b 1 )+log 10 (e i ) FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD 'log 10^ (Y i )=b 0 +b Ole  CompObj|~ jObjInfo Equation Native !1 X i 'Y ^  i =(10 b 0 )(10 b 1 ) X i FMicrosoft Equation 2.0 DS EquationEquation.COEE29qMD Y i -Y _1117612046FA>A>Ole CompObjjObjInfoEquation Native i_1117612150{F@cD>@cD>Ole CompObjji-1 i=2,K,n FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD (Y i -Y i-1 )-(Y i-1 -Y i-2 )=Y i -2Y i-1 +YObjInfoEquation Native _1117612320F?F> F>Ole ! i-2 i=3,K,n FMicrosoft Equation 2.0 DS EquationEquation.COEE29q{D Y i -Y i-1 Y i-1 100%i=2,K,nCompObj"jObjInfo$Equation Native %_1117614958FBH>BH>Ole (CompObj)jObjInfo+Equation Native , FMicrosoft Equation 2.0 DS EquationEquation.COEE29qgD Y i =A 0 +A 1 Y i-1 +e i_1117615060F 0I> 0I>Ole /CompObj0jObjInfo2 FMicrosoft Equation 2.0 DS EquationEquation.COEE29qcD 'Y ^  i =a 0 +a 1 Y i-1 FMicrosoft Equation 2.0 DS EquationEquation.COEE29qEquation Native 3_1117615190Fw K>w K>Ole 5CompObj6jObjInfo8Equation Native 9_1117615234F&L>&L>Ole <D Y i =A 0 +A 1 Y i-1 +L+A P Y i-P +e i FMicrosoft Equation 2.0 DS EquationEquation.COEE29qCompObj=jObjInfo?Equation Native @_1117617489%FHFN>HFN>D 'Y ^  i =a 0 +a 1 Y i-1 +L+a P Y i-P FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD 'Y ^Ole CCompObjDjObjInfoFEquation Native G6_1117617590F0Q P>0Q P>Ole HCompObjIjObjInfoK FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD jP FMicrosoft Equation 2.0 DS EquationEquation.COEE29qEquation Native L._1117617574FuQ>uQ>Ole MCompObjNjObjInfoPEquation Native Q._1117617918/FR>R>Ole RD j>P FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD e i =Y i -'Y ^  i MAD=CompObjSjObjInfoUEquation Native V_1117619996F`QT>`QT>|Y i -'Y i^ | i=1n  n=|e i | i=1n  nMSE=(Y i -'Y i^ ) 2i=1n  n=e i2i=1n  n FMicrosoft Equation 2.0 DS EqOle ^CompObj_jObjInfoaEquation Native b{uationEquation.COEE29q_D Y i =b 0 b 1X i b 2Q 1 b 3Q 2 b 4Q 3 e i log 10 (Y i )=log 10 (b 0 )+X i log 10 (b 1 )+Q 1 log 10 (b 2 )+Q 2 log 10 (b 3 )+Q 3 log 10 (b 4 )+log 10 (e i ) FMicrosoft Equation 2.0 DS EquationEquation.COEE29q_11176283134FࡹU>ࡹU>Ole lCompObjmjObjInfooEquation Native p_1116584923FkW>kW>Ole {CompObj|jtD Y i =b 0 b 1X i b 2M 1 b 3M 2 Lb 12M 11 e i log 10 (Y i )=log 10 (b 0 )+X i log 10 (b 1 )+M 1 log 10 (b 2 )+M 2 log 10 (b 3 )+L+M 11 log 10 (b 12 )+log 10 (e i ) FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD ObjInfo~Equation Native #_1117102536F`Y>`Y>Ole  FMicrosoft Equation 2.0 DS EquationEquation.COEE29q+D H 0 :b 1 =0H A :b 1 0TS:F obs =MSRMSE=SSR/kSSE/(n-k-1)RCompObjjObjInfoEquation Native G_1117103460F{Z>{Z>R:F obs F a,k,n-k-1 FMicrosoft Equation 2.0 DS EquationEquation.COEE29q`D b 1 t a2,n-2 S b 1 Ole CompObjjObjInfoEquation Native |_1117184393F0[>2<\>Ole CompObjjObjInfo FMicrosoft Equation 2.0 DS EquationEquation.COEE29qHD Y i =nb 0 +b 1i=1n  X ii=1n  X i Y ii=1n  =b 0 XEquation Native d_1117185055F^>^>Ole CompObjj ii=1n  +b 1 X i2i=1n  FMicrosoft Equation 2.0 DS EquationEquation.COEE29qTD b 1 =SSXYSSXXSSXY=(X i      !"T&%(')*+,-./0123546789:;<>?@ACBDEFHGIJKLNMOQPSRvUVWXYZ[\]^_`abcdefghijklmnopqrstuwz{|}~B  S A? 2Z3UeD ˒mQ1 `!Z3UeD ˒m(`@ :xcdd``a 2 ĜL0##4Y=F (.ÔCRcgb @x3PT obIFHeA*$/i ĺEA,a K353j&br]a|zVHA!hH!,\rZ'tC Jm2w-0WSV tB׉R=wBU~t'#1^jWjH\>Ri9;Y9ל:gJ 7;[@tLX-V39dj;cUN:0b@h٠d%BdQ `U:o5-O yTԾ{)9kٞV>QD VZ]s%[7y%'ǻNNr'Fu'M؉Ƿ1u؉O p7[m{>+~nld'6t e(1]w⟷JYX7.o1o kWd6BɠOєJG*#_}8>YF%*F6*F}Z\gI9`:#)9`:#.g:Y]j;O”]ʪ[]e5v=":);3e.J&%gUriQ౬r~1Yk}r\W!a0 mS}O#EBD %VY +DdhB  S A? 2l:0R4nHp`!l:0R4nHp` @|xcdd``> @c112BYLGcb n"&,0 ɲLj|QO*/%jg + "+@"yH"A"ٕpo &+ssU\|q#ܿϙ@Fl o ?pp/`7VVEl8`"v1bE$q8 k=@fL^ J H.6 M\ "v0o8221)Wx\ ]@0΃  /MDd$B  S A? 2[tN\!*`![tN\!*(@8z/YxTMhA~36Mbl1MCCQ4Ń i"Tr [9q$HoūBz*UD{3w\5a6o|)2#P!yQT 2x2g`Jql& 88?8ۤm H1;|<2Yt kU+{玨0ͫBV:3aymQu_t L6>\y.rB7ahL˾B9Mȶ"Y6^eDO̦`sp=j7CRu^V g*g+~߫^g H QK3bmt^"%̑yAǩ(E*g{osjZeUU ^vkokНQ/+ezE'pjκj<Kv-e8RC>hQͨ!kCzwj4-R["gzPK"bO9i(!QCTzQqśУ$:GBMo3b"nDX.,;|(dք]uWDdB  S A? 2 %Z;`(ű~a`!%Z;`(ű~8&@;xUMhA~3dLLM*ЋQh=49(BbPDo=՛`OB)xQ^ )U{M/x"D#5!&ԃx`5pOl?3ofK *f%DoѯģlӞ`kK*\va-~AݰmzgG#D FRMZ E%dD`m Ap!7]TTG،\,k I#z 4=\9Z`N9Ȱ%e2L푍2Av k}ZE5mDվ2xWy4ZiΩH~N&L>3 vx;|*~Z?~0fѻb(>rijdȧrLzkdKʺqaLJ۰j bdqS!FFQ,s1Hq'ϋZ3ْjp̍63.My"j],)Y.;F-$M?ԛU$7\/.fU ^ثu^9!dfɐw՘֙ɖ4SI1)mǪi?&(&G)'.H4>G7x$l)AVE r_bj΍ULԙ(ҊÜGNJ.L8w@ ;+KRsA<.E.b0΃ÕU紭DdB  S A? 2 J3` &`! J3` & Pxtxcdd`` @c112BYLGcby Amc }f:XX81Yd! 8Pk^pXtnS 4^ \$[YO]4|hg-sOoյea7@ (u+B"gȋ$<2d?Ȓ'-tVc frdJA./VGr]+\)K6*,|$H>SkǀVY7RW$8~FFH. DiCW3] (91t5.j @c112BYLGcbʯ6UT5-}}9͌oU k>~d=))K%CA1aRl% „`?+^MrU3r0 {1[ɼ &ۭ$"ݘa2TsGqj{(m_SsfP@%Ydo&`"wX(|F{0Y320C*/ڿX ?k5 ɇ5bUkUbT{y n0_k < KQ-Iު2GTf^(z/7'q hLP72b^ ˯Vxy_aDd @B  S A? 2!=f=ḳ")+`!=f=ḳ")b! xK@]äՆB ׊"RE muuAAMDpT\Eq7N*(ƻk.ӡ8'w9*z=~:T,"!l۶v%(0lţ4Dl thF=Nc*rr0P[*U`Sy=k ^(a&g ^6LumGR4;LVDch2`A}1$HLDq){BܤW#.W%.U=i:8ՑlMJd%]! 2YDt0' c'#7.񐻐L^tAw~QSD9?S^JuDU8r,04Yu+#M&aC#h=ml!W."dfJ>h%)g S~U4тDdB  S A? 2g-#B#Qa`D`>C `!;-#B#Qa`D`>Hh xcdd``fbd``baV d,F1J@Yp1B@?6 FsC0&dT20 rosv Й! ~ Ay ۏ/? N300}`ĵфr*Dd?Mf` 69d 1j`* I=bdbR ,.IAď ;/G3X?YaDdhB  S A? 2gQ-"beݩC `!;Q-"beݩ@H| xcdd``fbd``baV d,F1J@Yp1&B@?6 F~bnĒʂT$@o].:S $37X/\!(?71aXr2]b00tUN?, L&T`@[!&3BMlQe4! 1b;ZLLJ% "CX1Xa`eb DdhB  S A? 2g|ʊr^{]C `!;|ʊr^{]@H| xcdd``fbd``baV d,F1J@Yp1&B@?6 F~bnĒʂT$@o].:S $37X/\!(?71aXr2]b00tUN?, L;&gW&c@[!&3BMlQe4! 1b;ZLLJ% "CX1Xa`bDd B  S A? 2ivYf0P3E `!=vYf0P3 L" xOHQ7aa CE< AXH!d1`䪸À"x๳BQg!H ͛@lg~~{ $ȕBƴ0 ֥V dܫζu) r/#I'HFh3o|!+:;.#p(rl,%iȡMd C~8Dj&ǖg'x ZWO.U=ou_e~FZ4˜Q+w;Z]iKi^0soќ6hB|'ɱu7$d-MFKDz%IOW{RTx&{9j%x2P}JYmiYe6g!0yMoDx9Iп?C{.Z4SՓonw$!v{V2[ӼiMУ pG~ OІ7Dd$@B  S A? 2H| mY!&qEy~}'`!uH| mY!&qEy~pHD Cxcdd```d``baV d,F1J@Yp1&lB@?6  l@6XnĒʂT$@o].:S $37X/\!(?71aXr2]b5@:#+E3DBDxX@"Ε" ;GF0ɬdFacEfe =`\Ph.0}# \؟P=adbR ,.IAď ;/G=3X?p}DDd@B  S A? 2?hX ا+gm>)`!?hX ا+gm4 Pxcdd``bd``baV d,F1J@Yp1&B@?6  ؁lP5< %! H6&]t@Hfnj_jBP~nbS  ,@e@@1q%0j0tFV^|g D+EJYA""@" P)&e`܅o]YiYɌ.Ɗ.bφ. 027)?(pWA\^m+/3pAc `Cm``cI)$5bPd+? 35,z|䆘|{@9 !Cl``s2#RpeqIj.ŠQ ? 3= \5ąL zNJ.΅.I\_ $a` 2 . $b$bN-Fp/ 0.h|s n``S7#RpeqIj.ŠV "~<fyx9z1%zDdB   S A ?  2轇JwEW{) `!轇JwEW{ xtxcdd``> @c112BYLGcb,ː~d `!\R>,ː~ &@;|*xcdd`` @c112BYLGcbt WP[Ge+#J;.V]p5j|@y9#!q3#dF6:85l.ܤr'kBc^l+/EE@"H"E0{> |ឥ\+L JfAFj t5"*0])udAe\D "\Py<,dhPkFt.eP5CS#T O =?D@Cn 3&Bt/h D\D4"0za$F̹rfNQw2DU^P86.p H8 0y{e#RpeqIj.ŠV "~<fyx9z:`JDdB   S A ?  2lE>Xp:'H `!@E>Xp:'`)xtxcdd`` @c112BYLGcb`!ˑd-tJ^uo xcdd`` @c112BYLGcbUMO>e|1b\3Hnt/tE"R0X|mA-&/+ss\/}&"l"\"Y| [$QAt5A m$.X\E'Wq.r]ě]0H8L=Q + "".4B)AW]R5 epI cxi-|C\+J n.nDl`b\ m `pĤ\Y\ q1(2tA ;/G3XDd B  S A? 2E;p'!C`!;p'ӊ @x?H#A߼zY"{ڨ'C4!hT47EJ Whgq Z X\q)EŽyOL{qg t]MbDȘay::ujY1/Q_s 4Hy "0&r; (V |p EEdjx2ʈ(=VCaٖUBX.mx%ߛZ-Ws Wg969[ɖI/D%i"+<Еy*C_ֈ%?Fr팢!Ɨ.4/ϯ.kM>RGƩ49xBۈ|zŭRn;EC>w@c|jCmZ}iNJD3U;-Ⱥ$2,0U=QƾU֋vޢ+K."10I`!B>ޢ+K."10 (xt~xcdd``> @c112BYLGcb/0ɰ0,|C$0.hj 0y{)I)$5bPd(v<^Wf,]ADd0 @B  S A? 2*gw}-gK`!gw}-g: xK1_Ҫk:8 (j]  V-An8 "(ܤSEP<%֠R=/ｗ |C #E;;z((YDho C/8lDܦQm&|y9G&@ +p_ ɭ̒1cZz!U˦ðo,$,A43 N#ITJ*mꪇ)vpߺl{yK2PI ٕ$V5U%4'gzΒ*Xuz$y32@5UDTuZH4[^ْ vxھVF~{?{OHA u%tHm7f5"ZӤԝ]jSts6>;a˒R,˪?jcDdP B   S A ? 2!Bߤ!YF$͉}MN`!Bߤ!YF$͉}zxtxKAg'%&X"H+XiD4()A8-R ( `lRll๻w-$j`_ν7&UQM^ K',] Ơ~ t/m*L4HZ-|`X:NM=ɦl6Y+g7v-vgB ř%qa'$N,Z$Trz jqҟY1!:dG*mWI9'UTTcFux_ 'B3w~wFyfW` \G沛؉$H 4?~N4\Izb>To-FiAg5v+I)cje&Nj -ͻn,i:݅qpK,i0BޟV/,DdB   S A ?  2Yݨ爟dh5Q`!-ݨ爟dh`!hxcdd`` @c112BYLGcbc0r``錬 5Q4DLd&`W ؿ! 0y{~bdbR ,.I ca, HLDdl,B   S A ?  2rpJ14S;'NR`!FpJ14S;'xcdd`` @c112BYLGcb QDdl,B   S A ?  2rh9m{㧛nNT`!Fh9m{㧛nxcdd`` @c112BYLGcb 8(QDdB   S A ?  22$D[ZIMŠw\sW`!$D[ZIMŠw\s@/ X@I~ x]HQϽ3k:n쪫[Ƹ~AAHATF Zː}KY(LH(DC"KD_{g9̞{ܙ!uv;(!nDh6E&H؍S3E59@ 6Q#3&w5~ /qM ;D4"Ŋ%[(l8"١d%I!s%|_KbM%ɴy?Nmvlɝַw&[^*2CcQ]N De& d@h,[i˼sKx\T*ڰp=TO7qq:kfUg}:lD{\ּdQ?%~M&B^zGK>xYŀrMaWS)vbj\[-s-JgC^ǰ-.]rOw;/:it;5ywm EztNV^M.Ě+8O`R#w*뺑kmG~=d>ȹRL>F1iK^6ɪ}_[iDgH$Y./͘$zz):jT0I:Fyak"!L&Øl`*lu֨"#u3;zszr TI JiM`M&.58_NC|]);:1_6BiJȀ 8/m<~2ƽׄ;hDd"8B  S A?  2SQ.u^Qs/Z`!'Q.u^Qs`7Uv xOHQ73V-WG_[ebn.),s :xEP(HCA͛y37{0e7y7vR @G!=%BrNva`CˡrAzm=^I:vN O^Sqq`GlbqMHhd|B=OgƓ×`bڠjNh@(L3.9KZ4lF28$Jɷ,pbŪ+a߫N g1m ;N,H|/L@-W5:ocؽ]so2{Z -®~}b24dh)3=WBbWrDd%8B   S A ? %2~Q [[U Z`!RQ [[U @;[v xMLAlaK CQ*4LDjhRS`x"kp=pЃ^4ACB.o{@l̯y;KzQ A W!<I#"eYkTbySNG r,3tf= jL/ $.@V`K{҈=~ kcڗfв{=(9(Q)V!T7O&"Q6S1^,'☈WDd@<  C A? 2hAd!6 ^_p`!hAd!6 ^_p Ƚ ixcdd``c!0 ĜL0##4Y=F (.2KA?H 6XnĒʂT.(X'#M`AyDd x<   C A ? 2&l 0JEM`!l 0JEMXX! xMhA&&`H Pm<b[5m!(ZC% T,# x(3.ERqa߾7~Yl9X0Bzym v ͒ϳ[nW CAS{,|$ M_WJ=5,=eՉd 1ߥnQPLQ!~&<JC*?Z,LN92YŏeVwcJ)( BCS潯@?jcnu[e 9/b0EYOUp+O[xq#~t#& .މaӫLM΅v|~|OrP}Dd< ! C A!? 2 R/ɺnRi*N`!R/ɺnRi*L`x :xSJAKr%`0"ˆR@(0p,* ;_BDL <粁D$ƅٛnfv-leAqzΒq *,YG@avǭq`%WNj YH1bF^yAO7'ֿE̩Ra\-Xz+]Ӳvs HY2 <2Omu`>Ole  -X)(Y i -Y)=X i Y ii=1n  -X ii=1n  ()Y ii=1n  ()n i=1n  SSXX=(X i -X) 2i=1n  =X i2i=1n  -X ii=1n  () 2 n FMicrosoft Equation 2.0 DS EquationEquation.COEE29qCD b 0 =Y-b 1 X FMicrosoft Equation 2.0 DS EqCompObjjObjInfoEquation Native __1117185348F b> b>Ole CompObjjObjInfoEquation Native uationEquation.COEE29qD SST=(Y i -Y) 2i=1n  =Y i2i=1n  -Y ii=1n  () 2 n_1117185631Fp-0d>p-0d>Ole CompObjjObjInfo FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD SSR='Y i^ -Y() i=1n   2 =(SSXY) 2 SSXX=b 0 Y ii=1n  +b 1 X i Y iEquation Native _1117185945Ff>f>Ole CompObjji=1n  -Y ii=1n  () 2 n FMicrosoft Equation 2.0 DS EquationEquation.COEE29qQD SSE=Y i -'Y ^  i () 2ObjInfoEquation Native m_1117191840Fpg>pg>Ole i=1n  =SST-SSR=Y i2i=1n  -b 0 Y ii=1n  -b 1 X i Y ii=1n  FMicrosoft Equation 2.0 DS EquationEquation.COEE29qCompObjjObjInfoEquation Native _1117190311F@~ni>@~ni>D S b 1  =S YX  SSXX  =S YX  (X i -X) 2i=1n   FMicrosoft Equation 2.0 DS EquationEquation.COEE29qOle CompObjjObjInfoEquation Native D 'Y ^  i =b 0 +b 1 X 1i +L+b k X ki FMicrosoft Equation 2.0 DS EquationEquation.COEE29qCD R 2 =r _1117190998Fk>k>Ole CompObjjObjInfoEquation Native __1117191937Fl>l>Ole CompObjjY1,K,k2 ='Y ^  i -Y() 2i=1n  Y i -Y() 2i=1n  =SSRSST=1-SSESST FMicrosoft Equation 2.0 DS EquationEquation.COEE29qObjInfoEquation Native @_1117468469FLqn>Lqn>Ole $D Adj-R 2 =r adj2 =1-n-1n-k-1()SSESST()=1-1-r Y1,...,k2 ()n-1n-k-1()[] FMicrosoft Equation 2.0 DS EqCompObjjObjInfoEquation Native _1117468560 Fp>p>uationEquation.COEE29qD Y i =b 0 +b 1 X 1i +b 2 X 2i +e i E(e i )=0 FMicrosoft Equation 2.0 DS EqOle CompObj jObjInfo Equation Native uationEquation.COEE29qD Y i =b 0 +b 1 X 1i +b 2 X 2i +b 3 X 1i X 2i +e i E(e i )=0_1117474958F}r> s>Ole CompObj jObjInfo  !"#&)*+,-03456789:;<?BCDEFGHKNOPQRSTVWXYZ[]^_`abce FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD X=log 10 (X)X=10 X X=ln(X)X=e X FMicrosoft Equation 2.0 DS EqEquation Native _1117474658F!u>!u>Ole CompObjjuationEquation.COEE29qD Y i =e b 0 +b 1 X i1 +b 2 X 2i e i FMicrosoft Equation 2.0 DS EquationEquation.COEE29qObjInfo Equation Native  _1117476601F4Rv>4Rv>Ole  CompObjjObjInfoEquation Native c_1117608170X F@ w>@ w>GD VIF j =11-r j2 FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD MA(3) i Ole CompObjjObjInfoEquation Native :=Y i-1 +Y i +Y i+1 3MA(5) i =Y i-2 +Y i-1 +Y i +Y i+1 +Y i+2 5 FMicrosoft Equation 2.0 DS EquationEquation.COEE29q_1117609595"F+.y>+.y>Ole CompObj!#jObjInfo$Equation Native  _1117615341*'Fpz>pz>Ole $CompObj&(%jD E 1 =Y 1 E i =WY i +(1-W)E i-1 i=2,3,K0<W<1 FMicrosoft Equation 2.0 DS EquationEquation.COEE29qObjInfo)'Equation Native (B_1117617003,F`gu|>`gu|>Ole .&D H 0 :A P =0H A :A P 0TS:t obs =a P S A P  RR:|t obs |t a2,n-2P-1 FMicrosoft Equation 2.0 DS EqCompObj+-/jObjInfo.1Equation Native 2_11176197421F01 ~>01 ~>uationEquation.COEE29q,D 'Y ^  n+1 =a 0 +a 1 Y n +a 2 Y n-1 L+a P Y n-P+1 'Y ^  n+2 =a 0 +a 1 'Y ^  n+1 +a 2 Y n L+a P Y n-P+2 'Y ^  n+j =a 0 +a 1 'Y ^  n+j-1 +a 2 'Y ^  n+j-2 L+a P 'Y ^  n+j-P FMicrosoft Equation 2.0 DS EquationEquation.COEE29qOle =CompObj02>jObjInfo3@Equation Native A,D 'log 10 (Y i ) ^ =b 0 +b 1 X i +b 2 Q 1 +b 3 Q 2 +b 4 Q 3 'Y ^  i =10 b 0 +b 1 X i +b 2 Q 1 +b 3 Q 2 +b 4 Q 3 FMicrosoft Equation 2.0 DS EquationEquation.COEE29qD 'log 10 (Y i ) ^ =b 0 +b 1 X i +b 2 M 1 +_11176284866F@F>@F>Ole ICompObj57JjObjInfo8LEquation Native M1Tabley;SummaryInformation(;UDocumentSummaryInformation8\b 3 M 2 +L+b 12 M 11 'Y ^  i =10 b 0 +b 1 X i +b 2 M 1 +b 3 M 2 +L+b 12 M 11Oh+'0 $ @ L Xdlt|X4x^?w7]=wg䁄l^iҔsf5*lW8&`369~. jv(bbѸzYWYyݐ 8E.>mwu&n׾_IɸygJʬ5ev+o4/yW9!ı.$lsQBѐ8G֭اU-ݫ!M k,d|0ϳsP5⋮}k&>ǚ0{f?Xsz͓;h4ݦB߸zђU &ۭz# WczVo':c%ϹDd < 1 C A-? 2)0QrR#b%`($`!0QrR#b%`( "x5Rx_hG;owws˟і4<h)EД<L±<DJ-(r zXćB)Ml|Zl3 uawg?74@.d?`.FQJj[SAf1[k>SHC n' ݴ5]{GYC;M`Wо(:$+%ҝ`]]bD5\vh\1u&)ؘeV1tl#Y|^Wg_xd@<7l/dvKS7 %Δih`>QN(_MwwOo J0cnMwj@ќ~5tLQFzFV566ddUߴx7*ͩy|_`nd*Q&mtzY#ԛ=ڴZzd?d_ʤ+;!|rO+d(%ThSX5 66ɪ><6qUC6R} 3(tGn"e5[N@LGN09bbdls14}:鐻A{䏰 sj{3JJ}V%͸/A} R޵P}E6_'dI\ U RyҘM>ߊ S |vi eΦ3!u˔4&M2鿆de.,!WLRt%b|. :2ʿ3 $] dҡudZB:L^Dw >u{lb7øgU cJ,t>pmXK$ӞpQOP !nkx?Ͼth^%k7 8 !LSDd(|< 2 C A.? 2"S)34KR8)`!"S)34KRʞ@` 0excdd``g 2 ĜL0##4Y=F (.Ô!d3H1icY+@<znĒʂT$/&`b]F"L : 0& &B~356g' L ! ~ Ay ۏ/? N30@md F ΤpfHdUH&+l WPw($w0p(0=Ǡ." (]L +ss8Lpbf[y.pA oA``けI)$5bPdFÏ ;/G+3X?ttKDdL< 3 C A/? 2yk!AQV騉+`!yk!AQV騉x$ ]xkAmw4K66ET*T zh‚-,zEAA="*I(I(M&3`d|o 1]5 'F(t:ڿpl ,V1Ϭ}fvBaYY.Ldj8<~F`em N-n%kSMNj|~" `pS ;lcr=^ "ּip0yQfT%-G]Rt: Y*y\0guӌ~f*eLSᑘ4ͫT%'*GKUWrDVWɪirlSUI nJnȓeZ f!(3p߲џFf:p"%9!kϚlB͘8T)C$ QIG%WFUE-Vi2)YΪNknB/mYy_0Έ߅PB Y% $/# )9ψ~n,=&}""=4k'>AէNW]4lw-M9e}wx/ ޳a/\/ ޸J׌&DdLL< 4 C A0? 2dAR?6KC:@.`!8AR?6KC:&C< xKQܙuG֟kilCeKK!.Aઠ IRP}ؠ*QzHPRv{ðϽ{ VHU;R\.'RuX (hlxyG\4ƞSGNX#q(S6yW^7q&ca,va?nRktug 'É!x6Ň?$ PpfXzD"ݐeREʟ1f%-Qm4ƴHѼ oe9niZ܉JW9qq:l2.#\ՖWղijUu✴~^9}DA|8");vhp"Vxa ns~bSֈ9'"B8D>Xm3ĉ&9DI#k*Fp{X,Kzv2 Es缪59{#k*FHNͽ/2l{SVzLDҋadb 3q{kFqȍJ|"9yw7^Btw~ytC,[/%d&w)# ]ɸT$i}>C*'O׈$eI/6DS暝kޗ&}{E2j b̄B'ӝ?Mʕ֨'emkDd8< 5 C A1? 2FY_ʵ {362`!FY_ʵ {36&@<}xMhA̤ UXӺjn؋'% z!VZ =kOE'/{XEAԓ(([!hMR4aٙo޼ `G_T(!^v'qYfnjRawyR->bӄX~\Ι)<AӢ o˼Lx\T5*fv/Ã']!e\ J3>.3}xyKAMwd椷Í!ų(1,Ȕ$65L2&I9OT] 5nRaKN4 Rvf8eGk)]OGk1]۫H8\t Nijr-ֻ\r)*}S;fuأuسA]d5e܍b6EKKW eČbr"I# >_r,˻SwX 2ែ& !8֌z5$ݥ9=5vEfk&{ېw}doY߄Rvf %3 J,vo}<|\`H8cDd $< 8 C A4? 2ˁLk+5t l56`!ˁLk+5t lHD ux=haǟ雴I*ԡR3XX"]phZHi]Eqr(s)8 EpJZ/|_AG;6FH[a4{h+~dCw[G.84%Gh퓂cs~pM"CC*w7!o8GubHrBxr<~J3` Ŷcc l|K+@dH'b>q51leѡ,FGϲ]{W`K-vjNZt`A;u[d/|owWލ^rq"N&2TT4ɷ#ٹRY18w5Y53"= XDdx < 6 C A2? 2 ͔KVTd .9`!͔KVTd .Xxtxcdd`` @c112BYLGcbc0r``錬 g$UaEcDTYSqh?ݜΤ6!*i/c{+"A`"C"Fm/ -@&W&0;sr#ίLL #l3/EDD:9Ex`"aYY1D䀉BZh(V%2@B8Q2' ?y-m `p>adbR ,.I ca, DdkO< 7 C A3? 2W/82< 63f34<`!+/82< 63f"5+0, xKLA}]*-aC=ĀD@ăp$`B(11S`yu".^<8cz !O$NCꟘJ@ Y&< T_>26~?S+"D*Z*5SdZ3v8ҺS\ 2-BtcnpE&02D$fK I[2yfM4b8Et2G&M1K[J*]u29_oiԱyj\3OB:LEM7\BM\) aLs5Fށ\'UOԂ'L+7v\qnk,nfAr}YGV8*xS =P`\=2Ƶa2<O+jࢹuqX %/tJI9J6()R1K"Ş{s9ZL; o5$:9):`:>YFb6:XPtkuў:=ͅRұHyMsSr8TOk{J9 뫛X3Z1׃obwm$6MQU.Pր"QzX^ܖab=ǽjFP?U=RO ";%@"FH"7%A"H"[eA"H"`\B& +ss\ąO9pZNtX^6]~.H83r $0X,2'!iRj c3|1&M$zp D&Eڥ`"0|@okEIQ"ħFĦ`#Aed^Bl``K(F&&\A ~<fyx9 hDd 8<  C A? 2l6LX?czZA FCe\6/MNvmL38IR ?Z{_ ,'c?'xnfadI|Ut=ǕA2JVxGy+vQX(I%)vUg̜fX q v,Lt12'ދhDaO_å*eOdB6^&ܳˤهffB:8w1eG&/\Ns5\lѮZakyGQy>y6d dcp1ŀc7Լf5|_UQ[nʶ޺dH"2I>IC^\ YGA$ ߽yXI{r$D5s52ɳN*G2^QLLv)>V&I_} >s3k0C×gӤ4NS똦+δteo1^wRKEO=G ;L>Sv2/ϏqJgJ=gDd<  C A? 2{G=P{`!{G=P ' xkA&Gj!*HִzSbO(jR &H z("^ԓ "*C$To/"Eu~&sښOoE(mO:#;Bq,tGa8SdM&phu1XH1TlY+B 4_x=-Nr4>RjW+Siȑn-6,N69g(`zA|u)tG(N_;Zڈ(>'ePgAg1-{Uq[*I@JQJr]5J.rs@*$)"_EzZJ{_h[ܳr?|4ī#]|7x>"3쉊(3FDwXNXo<a1m=d|xT֌(9f A˚+ YswYNjlmD77ll}ɚeQ/YQx5TSc=25IW# Ys0,kޚADh]nh5^{a:sY@#O<~^k# ̟])}2Z3n_kFnunw|KYd3dr>`39AcW; aМ*ʀpϏ xӐ}UXfJDdp< ' C A'? 2:<:ӳլ+RK`!<:ӳլ+RKj - xkQϽifiG-DZ1L,Qb AMHE6r#(A R!ҍB\iэ*څT0ܙ.JlM̝_s=; 8aI7#9BlJQ9궘]^2R4 tr]##zUcjJb-dn~ce$r1Z;qrߒpOl9}ɌGDi&sߒ`}L m7Jkl^4>XVɘaELX'}掟G5GȚd@ٮ ,sCUH3L'i>uw;Mqƨ u2N{E{_/ʮ^w2xhy+ v}&?›FI' w' Chapter 11 Simple linear regression Cha Larry Winner SiarrNormali Larry Winner Si2rrMicrosoft Word 8.0e@F#@0ǃ>@0ǃ> _Q՜.+,D՜.+,` hp  UF Statistics )d  ' Chapter 11 Simple linear regression Title 6> _PID_GUIDAN{480EFDD7-B842-46C4-9AA8-F6C81FC47E98}  FMicrosoft Word Document MSWordDocWord.Document.89q [$@$NormalmH 6`6 Heading 1$@& 5>*CJ4`4 Heading 2$@&5CJ0`0 Heading 3$@&CJ<`< Heading 4$@&5CJ OJQJ8`8 Heading 5$@& CJ OJQJ4`4 Heading 6$@&5CJ@`@ Heading 7 $$@&5CJ$OJQJ<A@<Default Paragraph Font*B`* Body TextCJ:Wt"?\y 'Da~f    :Wt"?\y 'Da~     f^           PD 3Mf& -61. Q  ~ >%)-28L: CE+JNU~]cfjmqBtuupvilnortxz|}I Dy`k!%&r,01:6:>BF1KNUZ_c*gjo*tupvjmqsuwy{~ ".<J\1jPvpvkpvAY[.0Umo)+6NPH`b   ! # & > @ C [ ] c{}  *,x{ Zrt$<>!%9%;%%%%H)`)b)))))))M,e,g,o,,,---- . ..,......f3~33333444444444566&6>6@6666666{999999:::;;;O>g>i>>>>@@@BBB{LLLLLL M"M$MDO\O^OOOOES]S_STTTTTT U%U'UGU_UaUUUUUUUW W"W9WQWSWkWWWYYYYYYYZZ&Z>Z@Z_\w\y\\\\\\\] ^^``a&c>c@cf:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::57:RTWoqt":<?WY\tvy "$'?AD\^ay{~::::::::::::::::::::::::` 2$l 0JEM^2$hAd!6 ^_p a2$0QrR#b%`(a2$R/ɺnRi*e2$?j2gVg2$Z,Dw?-8j2$"S)34KRel2$yk!AQV騉n2$AR?6KC:@p2$FY_ʵ {36s2$͔KVTd .v2$9b#}E%_zx2$/82< 63f3{2$ˁLk+5t l 2$ 43PGQ?2$<:ӳլ+RK2$'C4T3}҆2$PAShlkXӈ2$"4"}Ί2$ MOI2$9Z-q{C|}ߓJ2$l6LX??BQRnZ.\aWcf@4  4H, H4L +4 4$4|44 H44 84  $4  c/4 4C4@4(`h4(T4Hh84 4h4}4D@4 844HJK^4h4 OLE_LINK2 OLE_LINK3 OLE_LINK1d eefd eef PA\1Up,6QHc  $ & A C ^ c~ -x{!Zu$?!%<%%%H)c)))))M,h,o,,---../...f333344444456&6A66666{9999::;;>>@ @BB{LLLL M%MDO_OOOES`STTTT U(UGUbUUUUUW#W9WTWkWWYYYYYZ&ZAZ_\z\\\\\]^___YcZcucwcff Larry Winner@C:\DOCUME~1\Owner\LOCALS~1\Temp\AutoRecovery save of mar5621.asd Larry Winner@C:\DOCUME~1\Owner\LOCALS~1\Temp\AutoRecovery save of mar5621.asd Larry Winner8C:\Documents and Settings\Owner\My Documents\mar5621.doc Larry Winner8C:\Documents and Settings\Owner\My Documents\mar5621.doc Larry Winner8C:\Documents and Settings\Owner\My Documents\mar5621.doc Larry Winner8C:\Documents and Settings\Owner\My Documents\mar5621.doc Larry Winner8C:\Documents and Settings\Owner\My Documents\mar5621.doc Larry Winner8C:\Documents and Settings\Owner\My Documents\mar5621.doc Larry Winner8C:\Documents and Settings\Owner\My Documents\mar5621.doc Larry WinnerA:\mar5621.doc0  >k w  $M<  V  =Z$ \$ [Q% i31P7  }:>ȄZ>; 87@ M mP NkS>ȄX>Ȅo4~b 1(g +[hVj0h K,.jBz&lm tVoK-o ~cr  hhOJQJo( hhOJQJo( hhOJQJo( hhOJQJo( hhOJQJo( hhOJQJo( hOJQJo(hh. hhOJQJo( hhOJQJo( hp@OJQJo( hhOJQJo( hOJQJo( hhOJQJo(hh. hhOJQJo( hhOJQJo( hOJQJo( hOJQJo( hhOJQJo( hhOJQJo( hp@OJQJo( hhOJQJo( hp@OJQJo(hh. hp@OJQJo( hhOJQJo( hhOJQJo( V87@1(g$M<[Q%w >k\$Z>;~cro4~bMP7+[hNkSX }:=K,.jtVo-omPi30h0 $&lm @tZy !+,-./23567f@@@@@ @D@@(@T@@4@l@@8@:@x@@@@@@D@F@@GzTimes New Roman5Symbol3& zArialG&Albertus Medium1 AlbaAVCharlesworth;Wingdings"1hQvQv_Q ) !0dd&& Chapter 11  Simple linear regression Larry Winner Larry WinnerCompObjdj