Data Mining Tools in SAS (SAS macros)
A set of SAS macros that support predictive modeling in terms of exploratory data analysis and model fit visualizations. A set of macros that help in the manipulation of strings (such as variable names) is also included such as:
- Add suffix to a list of variable names to be used to name new transformed variables.
- Remove certain variables from a list of variable names (e.g. remove variables from a model).
- Create strings representing variable interactions.
The full list of macros grouped by topic or area of application is shown below:
(An asterisk (*) next to the macro name indicates that help is available for the macro).
Use %Help() for a quick help on any of the macros marked with '*'.
Ex: %Help(PlotBinned)
%Categorize*: Categorize a set of variables in a given number of groups.
%Center*: Center and optionally standardize a set of variables.
%CheckRepeated: Check the existence of observations with repeated key values in a dataset
and creates an output dataset containing those observations.
%CreateGroupVar*: Create a variable identifying groups.
%CreateInteractions*: Create interactions between variables or interaction strings.
%CreateLags: Create a set of lagged variables with lag 1 to n.
%CreatePrevPostVar: Create a variable containing previous and posterior values of variables,
where the 'previous' and 'posterior' conditions are defined by a certain
matching criterion.
%DataComputeWithMissing: Generate data statements to handle missing values in a computed variable.
%Drop: Drop a set of variables in a dataset, regardless if they exist or not.
%GetVarList: Get the list of variables in a dataset associated with more complex listing
(like x1-x9, NUMERIC, etc.).
%GetVarNames: Get the names of the variables present in a dataset.
%GetVarOrder: Get the order in which variables are stored in a dataset.
%GetVarType: Get the type of a variable (C(haracter) or N(umeric)) in a dataset.
%Merge*: Merge 2 datasets without having them sorted by the by variables.
%SetVarOrder: Set the order in which variables are stored in a dataset.
%Standardize*: Center and standardize variables.
%Sort*: User friendly PROC SORT. Several datasets can be sorted simultaneously.
%Subset: Subset a dataset with a WHERE condition by by variables.
%Transpose*: Transpose a dataset.
%Boxcox*: Box-Cox transformation (for distribution normalization).
%Cov: Compute covariance matrix of a set of variables.
%CutMahalanobisChi: Detect multivariate outliers using Mahalanobis distance.
%Det: Compute determinant of a matrix.
%DetectOutliersHadi: Detect univariate outliers using the Box-Cox transformation and the Hadi algorithm.
%DetectOutliersMAD*: Detect univariate outliers based on the MAD.
%Hadi*: Robustly detect multivariate oultiers using Hadi method.
%Mahalanobis: Compute Mahalanobis distance of a set of variables.
%PHTestPlot*: Make plots of survival curves to visually check proportional hazards assumption.
%Qqplot: Make Q-Q plots for a set of variables.
%Aggregate: Aggregate data on a set of BY variables computing mean and mode.
%FreqMult*: Compute and store the frequencies of a list of variables in an output
dataset.
%GetStat: Get the value of ONE specified statistic for a set of variables
in a dataset.
%Means*: Execute a user friendly PROC MEANS.
%Colors: Create list of 10 color names to be used for graph symbols and lines.
%DefineSymbols: Define plotting symbols.
%GraphXY: Make a 2D plot of Y vs. X
%MPlot: Make multiple plots in the same window.
%PlotBinned*: Make 2D scatter or bubble plots of a target variable vs. binned continuous variables.
%Scatter: Make 2D scatter plots among pairwise variables.
%SetAxis: Create the string for an AXIS statement so that pretty values are shown.
%SymmetricAxis: Create the string for an AXIS statement so that the axis
is symmetric around 0.
%DetectCollinearities*: Detect collinearities among variables.
%Dfbetas*: Detect influential observations with the DFBETA criterion.
%EvaluationChart*: Plot an evaluation chart for a model with a categorical target.
%GainsChart*: Plot the Gains Chart for a model with a binary target.
%InformationValue*: Compute the Information Value of a set of variables for a binary target.
%KS*: Compute the KS statistic of a set of variables w.r.t. a binary target.
%LiftChart*: Plot the Lift Chart for a model with a discrete target.
%LogisticRegression: Perform a Logistic Regression iteratively by eliminating influential
observations.
%LorenzCurve: Plot the Lorenz Curve for a set of variables w.r.t. a non-negative target.
%PartialPlots: Make partial plots to graphically assess the linearity of variables in a
linear regression.
%PiecewiseTransf: Perform a linear piecewise transformation on continuous variables.
%QualifyVars: Automatically qualify variables into categorical or continuous.
%TestLogisticFit*: Test the fit of a logistic regression model.
%VariableImpact*: Make plots to evaluate the univariate impact of a set of continuous variable
on a binary target.
%Compute: Evaluate a given function by forcing the evaluation of its argument.
%EvalExp: Force the evaluation of an expression.
%ExistData: Check if a dataset exists.
%ExistMacroVar: Check if a macro variable exists.
%ExistVar: Check if all variables in a given list exist in a dataset.
%Getnobs: Get the nro. of observations and variables in a dataset.
%Help: Show this help information.
%All: Establish a condition that ALL variables in a list must satisfy.
%Any: Establish a condition that AT LEAST ONE variable in a list must satisfy.
%CompareLists: Compare two lists of names, and creates 3 datasets containing the list
of the names found in both lists, of the names found in one list and not
in the other, and viceversa.
%CreateVarFromList: Create a dataset with a column containing the names in a list.
%CreateVarList: Create a dataset with a column containing the variable names passed in a
list that are present in a dataset.
%FindInList: Find a set of names in a list and returns their word position.
%FindMatch: Find the names in a list that contain a given keyword.
%GetNroElements: Get the nro. of elements in a list.
%InsertInList: Insert a name in a list at a given position.
%MakeList: Make a list by adding a prefix and suffix to each name in a given list.
%MakeListFromName: Make a list from a name used as root.
%MakeListFromVar: Make a list containing the values of a variable in a dataset.
%MakeVar: Define global macro variables whose values are the elements of a given list.
%PrintNameList: Print a list of names in one column in the output window.
%RemoveFromList: Remove a set of names from a given list.
%RemoveRepeated: Remove repeated names from a given list.
%Rep: Generate a list of one value repeated a number of times.
%SelectName(s): Return a subset of names from a list.
%SelectVar: Return a list of variable names in a dataset matching a keyword.