Glossary

analytical model

a statistical model that is designed to perform a specific task or to predict the probability of a specific event.

attribute

See variable attribute.

backtesting

a procedure for monitoring the quality of behavioral and application scoring models. Backtesting validates the accuracy of the model's predictions.

baseline

the initial performance prediction against which the output data from later tasks is compared.

bin

a grouping of predictor variable values that is used for frequency analysis.

candidate model

a predictive model that evaluates a model's predictive power as compared with the champion model's predictive power.

challenger model

a model that is compared and assessed against a champion model for the purpose of replacing the champion model in a production scoring environment.

champion model

the best predictive model that is chosen from a pool of candidate models in a data mining environment.

characteristic report

a report that detects and quantifies shifts in the distribution of input variables over time in data that is used to create predictive models.

classification model

a predictive model that has a categorical, ordinal, or binary target.

clustering model

a model in which data sets are divided into mutually exclusive groups in such a way that the observations for each group are as close as possible to one another, and different groups are as far as possible from one another.

component file

a file that defines a predictive model. Component files can be SAS programs or data sets, XML files, log files, SPK files, or CSV files.

data model training

the process of building a predictive model from data.

data object

an object that holds the business data that is required to execute workflow tasks.

data set

See SAS data set.

data source (source)

a table, view, or file from which you will extract information. Sources can be in any format that SAS can access, on any supported hardware platform. The metadata for a source is typically an input to a job.

DATA step

in a SAS program, a group of statements that begins with a DATA statement and that ends with either a RUN statement, another DATA statement, a PROC statement, or the end of the job. The DATA step enables you to read raw data or other SAS data sets and to create SAS data sets.

DATA step fragment

a block of SAS code that does not begin with a DATA statement. In SAS Model Manager, all SAS Enterprise Miner models use DATA step fragments in their score code.

delta report

a report that compares the input and output variable attributes for each of the variables that are used to score two candidate models.

dynamic lift report

a graphical report that plots the sequential lift performance of one or more models over time, against test data.

file reference

See fileref.

fileref (file reference)

a name that is temporarily assigned to an external file or to an aggregate storage location such as a directory or a folder. The fileref identifies the file or the storage location to SAS. See also libref.

format

See SAS format.

Gini coefficient

a benchmark statistic that is a measure of the inequality of distribution, and that can be used to summarize the predictive accuracy of a model.

holdout data

a portion of the historical data that is set aside during model development. Holdout data can be used as test data to benchmark the fit and accuracy of the emerging predictive model.

identity

See metadata identity.

index

See SAS index.

informat

See SAS informat.

inner join

a join between two tables that returns all of the rows in one table that have one or more matching rows in the other table.

input variable

a variable that is used in a data mining process to predict the value of one or more target variables.

Kolmogorov-Smirnov chart

a chart that shows the measurement of the maximum vertical separation, or deviation between the cumulative distributions of events and non-events.

library reference

See libref.

libref (library reference)

a SAS name that is associated with the location of a SAS library. For example, in the name MYLIB.MYFILE, MYLIB is the libref, and MYFILE is a file in the SAS library.

life cycle phase

a collection of milestones that complete a major step in the process of selecting and monitoring a champion model. Typical life cycle phases include development, test, production, and retire.

logistic regression

a form of regression analysis in which the target variable (response variable) represents a binary-level, categorical, or ordinal-level response.

macro variable (symbolic variable)

a variable that is part of the SAS macro programming language. The value of a macro variable is a string that remains constant until you change it.

metadata

descriptive data about data that is stored and managed in a database, in order to facilitate access to captured and archived data for further use.

metadata identity (identity)

a metadata object that represents an individual user or a group of users in a SAS metadata environment. Each individual and group that accesses secured resources on a SAS Metadata Server should have a unique metadata identity within that server.

milestone

a collection of tasks that complete a significant event. The significant event can occur either in the process of selecting a champion model, or in the process of monitoring a champion model that is in a production environment.

model assessment

the process of determining how well a model predicts an outcome.

model function

the type of statistical model, such as classification, prediction, or segmentation.

model input variable report

reports the frequencies that input variables are used in the models for an organizational folder, a project, or a version.

model profile report

reports the profile data that is associated with the model input variables, output variables, and target variables.

model scoring (scoring)

the process of applying a model to new data in order to compute outputs.

model target variable report

a report that indicates the frequency in which target variables are used in the models that exist in the selected folder.

monitoring report

a report that consists of assessment charts, a ROC chart, a Gini Trend chart, a KS (Kolmogorov-Smirnov) chart, and a KS trend chart that can be used to compare the model performance curves of several candidate models.

neural network

any of a class of models that usually consist of a large number of neurons, interconnected in complex ways and organized into layers. Examples are flexible nonlinear regression models, discriminant models, data reduction models, and nonlinear dynamic systems.

observation

a row in a SAS data set. All of the data values in an observation are associated with a single entity such as a customer or a state. Each observation contains either one data value or a missing-value indicator for each variable.

package

See SAS package.

participant

a user, group, or role that is assigned to a task. These users, groups, and roles are defined in SAS metadata and are mapped to standard roles for the workflow.

performance table

a table that contains response data that is collected over a period of time. Performance tables are used to monitor the performance of a champion model that is in production.

PFD

See process flow diagram.

PMML

See Predictive Modeling Markup Language.

policy

a workflow element that associates event-driven logic with a task or subflow. Policies are usually triggered automatically by an event such as a status change or a timer event.

prediction model

a model that predicts the outcome of an interval target.

Predictive Modeling Markup Language (PMML)

an XML based standard for representing data mining results for scoring purposes. It enables the sharing and deployment of data mining results between applications and across data management systems.

process flow diagram (PFD)

a graphical sequence of interconnected symbols that represent an ordered set of steps or tasks that, when combined, form a workflow designed to yield an analytical result.

production models aging report

reports the number and the aging distribution of champion models.

profile data

information that consists of the model name, type, length, label, format, level, and role.

project

a collection of models, SAS programs, data tables, scoring tests, performance data, and reporting documents.

project tree

a hierarchical structure made up of folders and nodes that are related to a single folder or node one level above it and to zero, one, or more folders or nodes one level below it.

property

any of the characteristics of a component that collectively determine the component's appearance and behavior. Examples of types of properties are attributes and methods.

publication channel (SAS publication channel)

an information repository that has been established using the SAS Publishing Framework and that can be used to publish information to users and applications. See also publish.

publish

to deliver electronic information to one or more destinations. These destinations can include message queues, publication channels, and so on.

Publishing Framework

a component of SAS Integration Technologies that enables both users and applications to publish SAS files (including data sets, catalogs, and database views), and other digital content to a variety of destinations. The Publishing Framework also provides tools that enable both users and applications to receive and process published information.

receiver operating characteristic (ROC)

the name given to a chart used in signal detection theory to plot the sensitivity, or true positive rate, against the false positive rate (1 − specificity, or 1 − true negative rate) of binary data values. An ROC chart is used to assess a model's predictive performance.

ROC

See receiver operating characteristic.

SAS code model

a SAS program or a DATA step fragment that computes output values from input values. An example of a SAS code model is the LOGISTIC procedure.

SAS Content Server

a server that stores digital content (such as documents, reports, and images) that is created and used by SAS client applications. To interact with the server, clients use WebDAV-based protocols for access, versioning, collaboration, security, and searching.

SAS data set (data set)

a file whose contents are in one of the native SAS file formats. There are two types of SAS data sets: SAS data files and SAS data views.

SAS format (format)

a type of SAS language element that is used to write or display data values according to the data type: numeric, character, date, time, or timestamp.

SAS index (index)

a component of a SAS data set that enables SAS to access observations in the SAS data set quickly and efficiently. The purpose of SAS indexes is to optimize WHERE-clause processing and to facilitate BY-group processing.

SAS informat (informat)

a type of SAS language element that is used to read data values according to the data's type: numeric, character, date, time, or timestamp.

SAS Metadata Repository

a container for metadata that is managed by the SAS Metadata Server. See also SAS Metadata Server.

SAS Metadata Server

a multi-user server that enables users to read metadata from or write metadata to one or more SAS Metadata Repositories.

SAS Model Manager repository

a location in the SAS Content Server where SAS Model Manager data is stored, organized, and maintained.

SAS package (package)

a container for data that has been generated or collected for delivery to consumers by the SAS Publishing Framework. Packages can contain SAS files, binary files, HTML files, URLs, text files, viewer files, and metadata.

SAS publication channel

See publication channel.

SAS variable (variable)

a column in a SAS data set or in a SAS data view. The data values for each variable describe a single characteristic for all observations (rows). See also macro variable.

scoring

See model scoring.

scoring function

a user-defined function that is created by the SAS Scoring Accelerator from a scoring model and that is deployed inside the database.

scoring input table

a table that contains the variables and data that are used as input in a scoring test.

scoring output table

a table that contains the output variables and data that result from performing a scoring test. Before executing a scoring test, the scoring output table defines the variables to keep as the scoring results.

scoring test

a workflow that executes a model's score code.

segmentation model

a model that identifies and forms segments, or clusters, of individual observations that are associated with an attribute of interest.

source

See data source.

stability report

a graphical report that detects and quantifies shifts in the distribution of output variables over time in data that is produced by a model.

subscriber

a recipient of information that is published to a SAS publication channel.

swimlane

a workflow diagram element that enables you to group tasks that are assigned to the same participant.

symbolic variable

See macro variable.

target event value

for binary models, the value of a target variable that a model attempts to predict. In SAS Model Manager, the target event value is a property of a model.

target variable

a variable whose values are known in one or more data sets that are available (in training data, for example) but whose values are unknown in one or more future data sets (in a score data set, for example). Data mining models use data from known variables to predict the values of target variables.

task

See workflow task.

task status

the outcome of a task in a workflow. The status of a task (for example, Started, Canceled, Approved) is typically used to trigger the next task.

test table

a SAS data set that is used as input to a model that tests the accuracy of a model's output.

training data

data that contains input values and target values that are used to train and build predictive models.

universally unique identifier (UUID)

a number that is used to uniquely identify information in distributed systems without significant central coordination. There are 32 hexadecimal characters in a UUID, and these are divided into five groups with hyphens between them as follows: 8-4-4-4-12. Altogether the 16-byte (128-bit) canonical UUID has 36 characters (32 alphanumeric characters and 4 hyphens). For example: 123e4567-e89b-12d3-a456-426655440000

user-defined report

a customized report. The customized report is a SAS program and its auxiliary files and is stored on the workspace server that is used by SAS Model manager. User-defined reports are accessible from the New Reports wizard.

UUID

See universally unique identifier.

variable

See SAS variable.

variable attribute (attribute)

any of the following characteristics that are associated with a particular variable: name, label, format, informat, data type, and length.

WebDAV server

an HTTP server that supports the collaborative authoring of documents that are located on the server. The server supports the locking of documents, so that multiple authors cannot make changes to a document at the same time. It also associates metadata with documents in order to facilitate searching. The SAS business intelligence applications use this type of server primarily as a report repository. Common WebDAV servers include the Apache HTTP Server (with its WebDAV modules enabled), Xythos Software's WebFile Server, and Microsoft Corporation's Internet Information Server (IIS).

workflow

a series of tasks, together with the participants and the logic that is required to execute the tasks. A workflow includes policies, status values, and data objects.

workflow definition

a workflow template that has been uploaded to the server and activated. Workflow definitions are used by the SAS Workflow Engine to create new workflow instances.

workflow instance

a workflow that is running in the SAS Workflow Engine. After a workflow template is uploaded to the server and activated, client applications can use the template to create and run a new copy of the workflow definition. Each new copy is a workflow instance.

workflow task (task)

a workflow element that associates executable logic with an event such as a status change or timer event.

workflow template

a model of a workflow that has been saved to an XML file.