Know How:BusinessIntelligence:SAS:Solutions:RiskDimensions
Hauptseite > Know How > Business Intelligence > SAS > Solutions > Risk Dimensions > Glossar
| SAS |
|
Inhaltsverzeichnis |
Allgemein
- Instrumententypen
- Ein Instrumenttyp fasst die Attribute eines bestimmten Finanzinstruments zusammen. Beispiele für derartige Instrumente sind: FX Forwards, Staatspapiere, Zinsswaps, Aktienoptionen, Versicherungsprodukte usw. Bei der Definition neuer Instrumenttypen steht dem Anwender das objektorientierte Designkonzept der Vererbung zur Verfügung. Somit wird das An-legen neuer Instrumentenarten stark vereinfacht und neue Finanzinstrumente können problemlos hinzugefügt werden. Ein wichtiger Bestandteil der Definition von Instrumententypen ist die Bewer-tungs-Methode (siehe Methoden-Programme), die eingesetzt wird, um jede Instanz dieses In-strumententyps zu bewerten.
- Methoden-Programme und Methoden-Bibliothek
- Methodenprogramme werden
o zur Instrument-Bewertung, o zur Risikofaktor-Transformation o oder zur Modifizierung von Eingangsvariablen eingesetzt. Jedes von Benutzer definierte Methodenprogramm wird in der SAS Risk Management Methoden-Bibliothek gespeichert und in der graphischen Benutzeroberfläche auf. Bei der Risiko-faktor-Transformation können beispielsweise Zinskurven oder implizite Volatilitäten aus Marktda-ten berechnet werden. Dabei können sowohl Funktionen, die selbst in SAS programmiert sind, als auch Funktionen externer Bibliotheken (z.B. C oder C++ Bibliotheken), aufgerufen werden (siehe Abschnitt Schnittstelle zu externen Funktionen).
- Portfolio-Dateien
- Portfolio-Dateien sind Objekte, die den schnellen Zugriff auf Transaktions- bzw. Positionsda-ten innerhalb SAS Risk Management erlauben. Das flexible Konzept der Portfolio Definition er-laubt dabei, Daten unterschiedlicher physischer Datenquellen als ein logisches Portfolio-Objekt zusammenzufassen.
SAS Risk Management 1
- Portfolio-Filter
- Mit Hilfe von Portfolio-Filtern lassen sich gezielt Untergruppen verschiedener Portfolios mittels definierter Bedingungen selektieren. Beispielsweise kann ein bestimmte Währung, ein bestimmter Geschäftsbereich (z.B. Aktiva und Passiva), o.ä. selektiert werden. Die Risiko-Kenngrößen wer-den dann ebenfalls für diese Portfolio-Untergruppen separat ausgewiesen.
- Marktdaten-Modell
- Ein Marktdaten-Modell beschreibt funktional die individuelle Verteilung der Werte eines Risikofak-tors. Der Schlüssel bei der Implementation eines Marktdaten-Modells ist zum einen die Wahl der am besten geeigneten statistischen Verteilung, um das Verhalten eines bestimmten Risikofaktors zu beschreiben. Zum anderen die möglichst genaue Schätzung der zugehörigen Modell-Parameter, indem diese Parameter an bereits beobachtete Marktsituationen angepasst werden (‚fitted‘ Modell). Der Benutzer kann somit z.B. Modelle wie Brown’sche Bewegung, GARCH-Modelle, Vasichek-Modelle oder Cox-Ingersoll-Ross-Modelle sehr einfach selbst entwickeln und zur Vorhersage zukünftiger Risikofaktorwerte verwenden (siehe Monte Carlo Simulation).
- Analyse
- Eine Analyse ist beispielsweise eine Sensitivitätsberechnung oder eine Value-at-Risk-Berechnung unter Verwendung unterschiedlichster Methoden (z.B. Delta-Normal VaR, Historische Simulation, Monte Carlo Simulation, siehe Abschnitt Risk Engine‘).
• Die Ergebnisse aller Risikoanalysen lassen sich in jeder beliebigen Aggregationsebene der vom Benutzer spezifizierten Klassifikations-Variablen (z.B. Portfolio-Hierarchien, Geschäftsbereiche, Währungen) ausweisen: o addierbare Werte, wie z.B. Sensitivitäten, werden durch Summenbildung aggregiert, o nicht addierbare Werte, wie z.B. das Value-at-Risk (VaR), werden auf jeder Ebene berech-net. • Projekte bilden das zentrale Konzept für den Anwender. Ein Projekt kann aus Portfolios, Markt-daten, Markt-Modellen, Klassifikations-Variablen, Analysen und Berichten bestehen. Ein Pro-jekt wird verwendet, um die unterschiedlichen bereits definierten Objekte zusammenzuführen, um die gewünschten Risiko-Messgrößen zu berechnen, und sie in Form von Berichten zur Verfügung zu stellen. • Als Ergebnisse können alle für ein Projekt berechneten Kenngrößen ausgegeben werden.
A
- ACT
- AMO
- Add-In für Microsoft Office
- Abfragen / Berichte (Query and Reporting)
- Ein einfach zu bedienendes Werkzeug, das auch ungeschulten Anwendern Ad-hoc- Abfragen von Datenbanken und die Ausgabe von Berichten ermöglicht.
- Analytische Anwendungen
- Schlüsselfertige Lösungen, die die Entscheidungsprozesse innerhalb eines Bereichs oder einer Abteilung unterstützen. [Top]
B
- BID
- SAS BI Dashboard
- Best of Breed
- Ein Business-Intelligence- System, das aus Komponenten verschiedener Softwareanbieter besteht. Diese Systeme erfordern meist eine komplexe Verwaltung, erschweren die Weiterentwicklung und begrenzen die Möglichkeit des Wissensaustauschs. [Top]
- Business-Intelligence-Plattform
- Ein unternehmensweites System, das offen und modular alle Daten nachvollziehbar von der Extraktion bis zu analytischen Anwendungen integriert. Mit einer einheitlichen Steuerungs- und Abfragesprache, einem System verknüpfter Metadaten und zentraler Administration erlaubt die BI-Plattform die Speicherung und den Austausch von Wissen innerhalb eines Unternehmens. Sie vereinfacht die Systemarchitektur und kann für jede Art von Business-Intelligence- Projekten eingesetzt werden.
- Business-Intelligence-Portal
- Eine Web- Anwendung, die allen Nutzern eines Unternehmens einen zentralen, sicheren und personalisierten Zugang zu Business- Intelligence-Informationen oder -Anwendungen bereitstellt. BI-Portale können auch in vorhandene Unternehmensportale integriert werden.
D
- DI
- Data Integration
- DIS
- Data Integration Sutdio
- DWH
- Data Warehouse
- Data Mart
- Eine Business-Intelligence-Datenbank, die auf die Anforderungen einer Abteilung oder eines Unternehmensbereichs zugeschnitten ist.
- Data Mining
- Ein Modellierungswerkzeug zur Vorhersage von Verläufen und Risiken.
- DSS (Decision Support System)
- Eine Umgebung, in der Business-Intelligence- Informationen gespeichert, strukturiert und abgefragt werden können. [Top]
E
- EIS
- Executive Information Systems
Ein Führungsinformationssystem ist eine Scorecard- basierte Umgebung, die die Darstellung von Daten und definierbaren strategischen Indikatoren in summarischer und grafischer Form ermöglicht.
- ETL
- Extract, Transform, Load
- Ein Werkzeug, mit dem die Extraktion von Daten – etwa aus anderen Programmen wie ERP-Anwendungen – festgelegt und ausgeführt werden kann. Diese können dann transformiert und in eine Business-Intelligence-Datenbank, ein Data Mart oder ein Data Warehouse geladen werden.
I
- IDP
- Information Delivery Portal
- IVD = IndicatorViewDefinition
- Dateierweiterung für View Definitionen im BI Dashboard.
- Liegen Standardmäßig unter C:\SAS\config\Lev1\AppData\SASBIDashboard4.2\viewDefs
K
M
- Metadaten
Enthalten Informationen über Daten mit betriebswirtschaftlichem und technologischem Inhalt – unter anderem Verwendungszweck, Berechtigungen, Herkunft und Format. [Top]
- Multithreading
Eine Softwaretechnik, die das Aufteilen einer Task in logische Sub- Tasks innerhalb eines Prozessortaktes ermöglicht. So können die Sub-Tasks in Multiprozessor-Umgebungen auf mehrere Prozessoren verteilt werden. [Top]
O
- OCS = OLAP Cube Studio
- OLAP = Online Analytical Processing
Eine Klasse von Werkzeugen zur Datenauswertung. OLAP-Tools ermöglichen die Selektion von Detail- und Summendaten sowie die grafische Darstellung aus verschiedenen Perspektiven, zum Beispiel als Zeitreihen, nach geografischen Gesichtspunkten oder nach Produktkategorien. Wenn OLAPTools auf eine multidimensionale Datenbank aufsetzen, spricht man von MOLAP, im Gegensatz zu ROLAP bei der Auswertung relationaler Datenbanken. [Top]
- OLAP Cubes
Spezielle multidimensionale Datenbanken, die der direkten Auswertung von hoch verdichteten Daten dienen,wobei die Daten auf beliebige Art miteinander verknüpft werden können. [Top]
P
- Parallelisierung
Eine Softwaretechnik, die die gleichzeitige, parallele Verarbeitung auf mehreren Prozessoren oder mehreren in einer Gruppe zusammengefassten Computern ermöglicht. [Top]
- Prognose
Die „vorhersagende“ Analyse dient dazu, mit mathematischen, statistischen und Data-Mining-Methoden Modelle zu erstellen, die Prognosen künftiger Entwicklungen ermöglichen. [Top]
R
- Repository
Eine Datenstruktur zum Speichern von Metadaten, auch Metadaten- Dictionary genannt. [Top]
S
- SAS = Statistical Analysis System
- SEM = SAS Enterprise Miner
- SMC = SAS Management Console
- Star Model
Ein logisches Datenmodell für Data Warehouses, in dem alle Kennzahlen in einer zentralen Faktentabelle gespeichert werden, die mit Dimensionstabellen verknüpft ist. Diese enthalten zum Beispiel Bezeichnungen oder Hierarchiestufen als einzelne Tabellenspalten. [Top]
- Steuerzentrale (Administration Console)
Die zentrale Administrationskonsole des Business-Intelligence-Systems ermöglicht das Management von Datenquellen, Anwendungen, Datenbanken und Sicherheitsfunktionen sowie Test-, Entwicklungs- und Anwendungsumgebungen.
T
- TCO (Total Cost of Ownership)
Ein Verfahren zum Ermitteln der Gesamtkosten einer Investition. Zum reinen Anschaffungspreis werden die Kosten für Implementation, Schulung, Betrieb und Wartung hinzugerechnet.
W
- WebDAV = Web-based Distributed Authoring and Versioning
- WRS = Web Report Studio
SAS Glossar
- Archive
In the Publishing Framework, a result set package that is compressed and saved to a directory file. The archive may also be cataloged in an LDAP directory. The archive contains the contents of a package and metadata that is necessary for extracting the contents. An archive is compressed using ZIP compression and is saved with an SPK extension. The archive is saved to the specified location and remains available to users until you delete the archive. An object server can retrieve the archive upon request. See also Delivery Transport. The data compression is accomplished using zlib, (c) 1995-1996 Jean-loup Gailly and Mark Adler.
- Base
In LDAP, the distinguished name (DN) to use as a starting place for searches. Searches can begin at any point in the directory tree. Beginning the search lower in the tree can dramatically reduce both the search time and spurious results. For example, Base might be specified in SAS options or Workspace Manager parameters as BaseDN or ldapBase.
- Channel
In the Publishing Framework, a method of delivering result set packages to a set of subscribers of that channel. See also Delivery Transport.
- Component Object Model (COM)
Microsoft's specification for developing objects. COM is a language-neutral binary interface specification for Windows objects and a set of run-time functions for instantiating them. With COM, it is possible to build an object using one language, and use that object in another language. Both OLE and ActiveX were developed using COM. See also COM/DCOM Client.
- COM/DCOM Client
A program that uses Microsoft's Component Object Model or Distributed Component Object Model to make requests to a server. COM/DCOM clients can be written in Visual Basic, C++, Perl, or other languages in the Windows environment.
- Common Object Request Broker Architecture (CORBA)
A standard for distributed object communication that is created by the Object Management Group. CORBA is the most widely used distributed object standard for connecting operating system platforms from multiple vendors.
- Distributed Component Object Model (DCOM)
Microsoft's specification for distributing and using COM objects across systems in a network. See also Component Object Model (COM) and COM/DCOM Client.
- Delivery Transport
In the Publishing Framework, the method of delivering a result set package to the consumer. Supported transports include archive, e-mail, message queue, and WebDAV. Although not a true transport, a channel also functions as a delivery mechanism. See also Archive, E-mail, Message Queue, WebDAV, and Channel.
- Directory Information Tree (DIT)
In LDAP, the entries that a directory server contains. These entries are stored in a hierarchical, tree-like structure.
- Directory Schema
In LDAP, the set of defined attributes and object classes that defines the content of acceptable entries within the directory server.
- Distinguished Name (DN)
A unique identifier of an entry in an LDAP Network Directory. In effect, it is the path to the object in the directory information tree.
- DIT
See Directory Information Tree.
- DN
See Distinguished Name.
- Domain Name System (DNS) Name
In TCP/IP, the identifier of the TCP/IP host using nomenclature that is meaningful to people. For example, AlpineAirways.com might be the DNS name for the Alpine Airways Web server. The DNS server resolves DNS names to TCP/IP host addresses. For example, the DNS server might resolve AlpineAirways.com to 192.168.145.6.
In the Publishing Framework, a common method for delivering a result set package to recipients whose identities are known to the publisher. See also Delivery Transport.
- Entry filter
See Package Entry Type Filter and MIME Type Filter.
- Filter
In the Publishing Framework, an abstraction that specifies criteria for selecting a result set package or specific package entries for a subsequent operation. Filters are useful in the following contexts: channel subscription properties, viewer file processing, stored processes, and LDAP directory searches. For example, a channel subscriber can define a name/value filter that controls the types of packages that are delivered. The filter keys upon the name/value pair that a publisher has attached to the package that describes the package's content. See also Name/Value Pair, Name/Value Filter, Package Entry Type Filter, and MIME Type Filter.
- Group
In the Publishing Framework, a collection of subscribers. Subscribing groups to channels facilitates subscription management, which allows the convenient delivery of information to group members who share a common interest. When a group is subscribed to a channel, each member of the group who has a defined subscriber entry in the Publishing Framework will receive the information that is published to that channel. Only the administrator has the authority to add members to or remove members from a group.
- Integrated Object Model (IOM)
The set of object-based interfaces that is surfaced by SAS software when it is run as an object server.
- Integrated Object Model (IOM) Server
A SAS object server that is launched to fulfill client requests for IOM services.
- IOM Bridge for COM
The software component of SAS Integration Technologies that is used (transparently) when a Windows client accesses an IOM Server on a platform other than Windows. The bridge allows developers to write applications to the Microsoft COM/DCOM specifications that access SAS servers, such as OS/390 and UNIX, that do not support the COM/DCOM specifications.
- IOM Bridge for Java
The software component of SAS Integration Technologies that is used (transparently) when a Java client accesses an IOM Server. The bridge allows developers to write applications using Java that access MVA SAS servers.
- IP Address
The Internet protocol address of a host.
- Java Development Toolkit (JDK)
A Java software development environment that is available from Sun Microsystems, Inc. It includes a Java Runtime Environment (JRE), a compiler, a debugger, and other tools for developing Java applets and applications.
- Java Runtime Environment (JRE)
An execution environment for Java applications and applets. The JRE consists of the Java Virtual Machine (JVM), the Java platform core classes, and supporting files.
- Java Virtual Machine (JVM)
A program that interprets your Java code so that it can be executed on the native operating system of your machine. The JVM can run on either the client or the server. The JVM is the key software component that enables your Java programs to be portable across platforms. A JVM is included with JDKs and JREs from Sun Microsystems, Inc., as well as with most Web browsers.
- LDAP (Lightweight Directory Access Protocol)
The protocol that is used by applications to store and retrieve information in a network directory (LDAP server). SAS Integration Technologies supports LDAP Version 2, which is specified by RFC 1777.
LDIF (LDAP Data Interchange Format) file An open standard in which LDAP network directories share data. The LDAP server administrator also uses files in this format to populate the LDAP server.
localhost A keyword that is used to identify the host on which you are executing. If the client uses a sasServer definition with a sasMachineDNSName of localhost, the client connects to the host on which the client is executing.
Logical Name The name that is used by the Object Spawner Daemon, Workspace Factory, and Workspace Manager that allows for the assignment of a name to Object Server functionality. It is also used by SAS when publishing packages to an LDAP URL.
Message Queue In application messaging, a place where one program can send messages that will be retrieved by another program. The two programs communicate asynchronously without any knowledge of the location of the other program or even whether the other program is running. See also Delivery Transport.
MIME Type Filter In the Publishing Framework, it determines which file entries of a specific MIME type are included in packages that are published to that subscriber. MIME is an acronym for Multipurpose Internet Mail Extensions. A MIME file type is identified when the publisher inserts the file entry into the package. Use of a MIME type filter is limited to subscribers who specify a message queue transport in their subscription properties. Examples of MIME types include application/zip, image/jpeg, or text/html. See also Filter.
Name/Value Pair In the Publishing Framework, an attribute and value that the publisher uses to describe the contents of the package or a package entry. For example, "type=production" might be a name/value pair that is used to describe the entire package contents as production data. Both the entire package and a selected package entry can be assigned one or more site-specific name/value pairs. Subscribers can use name/value pairs as keys upon which to define name/value filters, which control the types of packages that are delivered to them. See also Name/Value Filter, Filter, and Viewer.
Name/Value Filter In the Publishing Framework, a subscription property that determines which packages are published to that subscriber. A subscriber filter is keyed to the name/value pair that the publisher specifies in the package. See also Name/Value Pair and Filter.
Namespace In relation to publishing to WebDAV-enabled servers, a URI that identifies an entry in a WebDAV package. Visit the W3C Web site for more information about namespaces in XML.
Network Directory Another name for an LDAP server. It is a directory that contains instances of object classes from a directory schema that are manipulated via the LDAP protocol.
Object Class In LDAP, a definition of a type of object that can be stored in the network directory. Each object class consists of a set of attributes. The attributes can be either mandatory or optional. The complete set of object classes for a directory defines it schema.
Object Management Group (OMG) A nonprofit industry consortium formed in April 1989 that develops standards for distributed objects. OMG standards include the Common Object Request Broker Architecture (CORBA) and the Internet Inter-ORB Protocol (IIOP).
Object Request Broker (ORB) In object-oriented programming, a piece of middleware that supports communication between clients and servers. When a client invokes a method that is supported by an object server class, the ORB finds an instance of the server class, invokes the requested method, and returns the results to the requesting client. An ORB allows clients and servers to dynamically discover the other and to communicate with each other across a network.
Object Server See IOM Server.
Object Spawner A process spawning service that instantiates object servers that are using the IOM Bridge protocol engine. In effect, it is a daemon on the server that listens for incoming client requests for IOM services. When the daemon receives a request from a new client, it launches an instance of an IOM server to fulfill the request. (Depending upon which incoming TCP/IP port the request was made, the spawner may alternatively initiate the administrator interface or service a request for a Universally Unique Identifier (UUID).)
OMG See Object Management Group.
ORB See Object Request Broker.
Overrides In the Publishing Framework, a set of subscription properties that are assigned to a single channel. These unique properties override the default properties that a subscriber assigns to all channel subscriptions.
Package See Result Set Package.
Package Filter See Name/Value Filter.
Package Entry Type Filter In the Publishing Framework, it determines the types of package entries to be published to that subscriber. Use of a package entry type filter is limited to subscribers who specify a message queue transport in their subscription properties. Examples of entry types include binary, HTML, and MDDB. See also Name/Value Pair and Filter.
Package Entry MIME Type Filter See MIME Type Filter.
Port A number that is used to direct traffic to a particular entity on a TCP/IP host. Servers listen on a specific port number, and the TCP/IP protocol stack on the host delivers traffic to each server based on this number.
Publisher In the Publishing Framework, any person, SAS program, or application that distributes information using the Publishing Framework.
Queue See Message Queue.
Result Set Package In the Publishing Framework, a container for content that was produced by a SAS System program execution or a third-party application. Package content takes these forms SAS file SAS catalog SAS data set SAS database (such as DMDB, FDB, and MDDB) SAS SQL view Binary file (such as Excel, GIF, JPG, PDF, PowerPoint, and Word) HTML file (including ODS output) Reference string (such as a URL) Text file (such as a SAS program) Viewer file (an HTML template that formats SAS file items for viewing in a view-only transport, such as e-mail).
Round-robin A scheduling algorithm that iterates through all available choices before beginning again with the first choice. For example, if there are three choices, A, B, and C, the scheduler would process A the first time, B the second, and C the third. Then the process would repeat starting with A and so on.
Schema See Directory Schema.
Service Name In TCP/IP, a port identifier. Service names are defined in your installation's services file. The location of the services file is host dependent, but it is typically located as follows: For OS/390: The location is site specific. Contact your network administrator. For UNIX: /etc/services For Windows NT: c:\winnt\system32\drivers\etc\services
Spawner Configuration File An LDIF file that contains the information that is needed to launch an Object Spawner Daemon.
Stubs Java classes that act as interfaces for remote software objects. Stubs are instantiated on the client and are passed method calls, which the stub then repackages for delivery through an ORB to remote software objects. In many distributed object systems, stubs are referred to as proxies.
Subscriber In the Publishing Framework, a recipient of information that is published to a channel.
Subscription In the Publishing Framework, the association of a subscriber or a group to a channel.
Subscription Manager Applet In the Publishing Framework, a Java applet that enables subscribers to create and manage their own subscriptions.
URI (Uniform Resource Identifier) It identifies entries in WebDAV packages. URIs are documented in RFC2396.
UUID (Universally Unique Identifier) Also known as a Globally Unique Identifier (GUID), a 128 bit identifier that is comprised of date/time information and a host's IEEE node address. Its 128 bits represent 1 x 10 * * 39 unique identifiers.
Viewer In the Publishing Framework, a custom-written template that contains HTML tags for formatting result set package content for view-only transports, such as e-mail. The viewer file is applied to the package during package publishing. A viewer file is particularly useful for rendering a SAS data set in a tabular format for viewing in e-mail. Also, it is useful for streaming dynamic information in the form of a binary file or a URL for inclusion in an electronic newsletter format.
WebDAV (Web Distributed Authoring and Versioning) An emerging industry standard that is based on extensions to HTTP 1.1 that enables collaborative development of files and collections of files on remote Web servers. WebDAV is documented in RFC2518. Visit the WebDAV site for more information. See also Delivery Transport.
Workspace A SAS Integrated Object Model (IOM) workspace represents a single session with the SAS System. It is also a root object in the IOM object hierarchy. In function, it is similar to the Application object, which is available in many Microsoft products.
XML (Extensible Markup Language) A subset of SGML that enables generic SGML to be served, received, and processed on the Web in the way that is now possible with HTML. Visit the XML site for more information.
ARM, Logging
- appender
- a named entity that represents a specific output destination for messages.
- Destinations include fixed files, rolling files, operating system facilities, and client applications. You can configure appenders by specifying thresholds, filters, log directories and filenames, pattern layouts, and other parameters that control how messages are written to the destination.
- filter
- a set of character strings or thresholds, or a combination of strings and thresholds that you specify. Log events are compared to the filter to determine whether they should be processed.
- level
- the diagnostic level that is associated with a log event. The levels, from lowest to highest, are TRACE, DEBUG, INFO, WARN, ERROR, and FATAL.
- log event
- an occurrence that is reported by a program for possible inclusion in a log.
- logger
- a named entity that identifies a message category. Loggers are named using a hierarchical system that enables you to configure logging at a broad or a fine-grained level.
- The logging facility includes a set of high-level loggers for SAS servers, including Audit, Admin, App, IOM, and Perf. Some loggers are subdivided into lower-level (child) loggers. For example, the Audit logger has descendant loggers called Audit.Meta and Audit.Authentication, and Audit.Meta has descendant loggers called Audit.Meta.Security and Audit.Meta.Mgmt. The Root logger is the highest-level logger and does not represent a specific message category.
- Loggers inherit settings from their higher-level (ancestor) loggers.
- logging configuration
- an XML file or a set of SAS program statements that determines how log events are processed. You use the logging configuration to assign thresholds to loggers, to configure appenders, and to specify which categories and levels of log events are to be written to each appender.
- If you perform a planned deployment, then the SAS Deployment Wizard provides default logging configuration files for your SAS servers.
- pattern layout
- a template that you create to format messages. The pattern layout identifies the types of data, the order of the data, and the format of the data that is generated in a log event and is delivered as output.
- threshold
- the lowest event level that is processed. Log events whose levels are below the threshold are ignored.
