The LOGSELECT Procedure

The "Model Information" table displays basic information about the model, such as the response variable, the frequency variable, the link function, and the distribution of the data that is assumed by the LOGSELECT procedure. For information about how PROC LOGSELECT determines the response distribution, see the section Response Distributions. If you specify the CODE statement, then the table displays the name of the variable in the scoring code that contains the classifications. If you do not also specify the PCATALL option, then the table also displays the name of the variable in the scoring code that contains the predicted probabilities.

The "Number of Observations" table displays the number of observations read from the input data table and the number of observations used in the analysis. If a FREQ statement is present, the table displays the sum of the frequencies that are read and used. If a WEIGHT statement is specified, the table displays the sum of that are read and used. If you specify a PARTITION statement, the table displays the values for each role.

The "Response Profile" table displays the ordered value from which the LOGSELECT procedure determines the probability being modeled as an event in a binary model and displays the ordering of categories in a multinomial model. For each response category level, the frequency used in the analysis is reported. If you also specify the PRIOR= option, the scaled prior probabilities are also displayed. You can affect the ordering of the response values by specifying response-options in the MODEL statement. For a binary model, the note that follows the "Response Profile" table indicates which outcome is modeled as the event. For an ordinal multinomial model, the note that follows the "Response Profile" table indicates how the ordered response levels are accumulated; for a nominal multinomial model, the note indicates the reference response level. If you specify a PARTITION statement, the table displays the values for each role.

The "Class Level Information" table lists the levels of every variable specified in the CLASS statement. You should check this information to make sure that the data are correct. You can adjust the order of the CLASS variable levels by using the ORDER= option in the CLASS statement. You can suppress the "Class Level Information" table completely or partially by using the NOCLPRINT= option in the PROC LOGSELECT statement.

If the classification variables use a nonsingular parameterization, the "Class Level Information" table also displays the reference value for each variable.

When you specify the SELECTION statement, the LOGSELECT procedure by default produces a series of tables that display information about the model selection. The "Selection Information" table informs you about the model selection method, selection and stop criteria, and other parameters that govern the selection. You can suppress this table by specifying DETAILS=NONE in the SELECTION statement.

When you specify the ITHIST option in the PROC LOGSELECT statement, the "Iteration History" table displays, for each iteration of the optimization, the number of function evaluations (including gradient and Hessian evaluations), the value of the objective function, the change in the objective function from the previous iteration, and the absolute value of the largest (projected) gradient element. The objective function that is used in the optimization in the LOGSELECT procedure is normalized by default to enable comparisons across data tables that have different sampling intensity. You can control normalization by specifying the NORMALIZE= option in the PROC LOGSELECT statement.

The convergence status table is a small ODS table that appears as a message that indicates whether the optimization succeeded and which convergence criterion was met. If the optimization fails, the message indicates the reason for the failure. If you save the convergence status table to an output data set, a numeric Status variable is added that enables you to assess convergence programmatically. The values of the Status variable indicate the following:

When you specify DETAILS=ALL or DETAILS=STEPS in the SELECTION statement, the LOGSELECT procedure produces the "Entry Candidates" or "Removal Candidates" table, which displays the effect names and values of the criterion used to select entering or departing effects at each step of the selection process. For each step, the effects are displayed in sorted order from best to worst of the selection criterion.

When you specify the SELECTION statement, the LOGSELECT procedure produces the "Selection Summary" table, which displays information about which effects were added to or removed from the model in the various steps of the model selection process. The statistic that led to the entry or removal decision is also displayed. You can request further details about the model selection steps by specifying DETAILS=STEPS or DETAILS=ALL in the SELECTION statement. You can suppress the display of the "Selection Summary" table by specifying DETAILS=NONE in the SELECTION statement.

If you specify the LASSO selection method, then this table displays information about which effect was added to the model, the number of effects in the model, the lambda value, and the information criteria.

When you specify the SELECTION statement, the LOGSELECT procedure produces a simple table that tells you why model selection stopped.

When you specify the SELECTION statement, the LOGSELECT procedure produces a simple table that tells you why the final model was selected.

When you specify the SELECTION statement, the LOGSELECT procedure produces a simple table that tells you which effects are in the final model.

The "Dimensions" table displays size measures that are derived from the model. It displays the number of columns in the current design matrix, the number of effects in the current design, the largest number of design columns associated with an effect, the rank of the matrix, and the number of parameters in the current model, including any scale parameters.

The "Global Test" table provides a likelihood ratio test for the hypothesis of whether the final model provides a better fit than a model without effects (an "intercept-only" model). This table is not produced when the NOSTDERR option is specified nor when LASSO selection is performed.

If you specify the NOINT option in the MODEL statement, the reference model is one where the linear predictor is 0 for all observations.

The "Fit Statistics" table displays a variety of likelihood-based measures of fit. The values that the "Fit Statistics" table displays are not based on a normalized log-likelihood function. If you specify a PARTITION statement or the PARTFIT option, the table displays the values for each role along with statistics for comparing the training, validation, and testing results.

For more information about the statistics displayed in this table, see the section Model Fit and Assessment Statistics.

The parameter estimates, their estimated (asymptotic) standard errors, and p-values for the hypothesis that the parameter is 0 are presented in the "Parameter Estimates" table. If you request confidence intervals by specifying the CLB option in the MODEL statement, confidence limits are produced for the estimates.

When you specify the COVB option in the PROC LOGSELECT statement, the LOGSELECT procedure displays the covariance matrix of the parameter estimates. The covariance matrix is computed as the inverse of the negative of the matrix of second derivatives of the log-likelihood function with respect to the model parameters (the Hessian matrix), evaluated at the parameter estimates.

When you specify the CORRB option in the PROC LOGSELECT statement, the LOGSELECT procedure displays the correlation matrix of the parameter estimates.

When you specify the TYPE3 option in the MODEL statement, the LOGSELECT procedure produces the "Model Analysis of Variance (Type III)" table. This table displays tests that all parameters for a particular effect are equal to zero. For more information, see the section Joint Tests and Type 3 Tests.

When you specify the ASSOCIATION option in the PROC LOGSELECT statement and have binary, binomial, or ordinal response data, the "Association Statistics" table displays the concordance index c (the area under the ROC curve, AUC), Somers’ D statistic (Gini’s coefficient), Goodman-Kruskal’s gamma statistic, and Kendall’s tau-a statistic, along with raw counts of pairs of observations. If you also specify a PARTITION statement, a row is displayed for each role. For more information, see the section Association Statistics.

The "Classification" table is displayed if you specify the CTABLE option without specifying an output data set. If you also specify a PARTITION statement, a table is displayed for each role. For more information, see the section Classification Table and ROC Curves.

When you specify the PCATALL option in the CODE statement, the LOGSELECT procedure produces the "Score Code Variables for Predicted Probability" table. For each level of the response variable, this table displays the corresponding variable in the scoring code that contains the predicted probability.

The "Timing" table displays the amount of time (in seconds) that PROC LOGSELECT required to perform different tasks in the analysis.

The OutputCasTables table is a special table that has information about each CAS table that is created during a CAS action execution. The information for each CAS table consists of the CAS table name, the caslib in which the table resides, and the number of columns and rows in the CAS table. Because this table is not a typical ODS table that contains analytical results, you cannot include it in the table-spec-list in the DISPLAYOUT statement.