Glossary

A

ABC classification: A method of classifying items in decreasing order of importance, such as annual dollar volume or your company's transaction history.

absolute error: Magnitude of forecast errors, actual less forecast values, without regard to sign.

accuracy measure: In Oracle Demand Planning, a statistical data element that estimates the accuracy of a comparison between two measures. Accuracy measures include MAD, MAPE, and RMSE.See also Mean Absolute Deviation (MAD), Mean Absolute Percent Error (MAPE), and Root Mean Square Error (RMSE).

across: The column position in a Demand Planning report or worksheet. Columns go across the page. See also column.

activity log: Demand Planning feature that enables planners to view a record of system activities.

ad hoc analysis: In Oracle Demand Planning, the process of analyzing data using a report or graph for which you specify the dimensions, dimension values, and layout. While viewing an ad hoc analysis, you can drill up or down on aggregate values. See also graph, report.

adjustment: In Oracle Demand Planning, refers to the manual modification of forecast values to account for outliers or incorrect measurements. You can adjust a single value or a range of values. You can enter comments regarding the adjustments.

aggregate: In Oracle Demand Planning, a predefined grouping of values along a hierarchy, or a user-defined grouping of values. For example, a predefined aggregate for Geography might total territories by city, state, region, and country. A user-defined aggregate for Geography might combine values for three cities that are of interest to the user. See also custom aggregate.

aggregate forecast: For a forecast variable such as demand, adding the contributions of child levels to obtain a single forecast value for the parent level.

aggregation: In Oracle Demand Planning, the consolidation of data for lower level dimension values into parent values. Data is often collected at the lowest available level of detail and is aggregated into higher level totals for analysis. For example, units sold in various cities might be aggregated into total units sold for all cities in a region. The aggregation rule might be “add,” in which case the values are added, or there might be an alternate aggregation rule such as “average.”

aggregation level: In Oracle Demand Planning, the position in a dimension hierarchy at which data can be viewed or manipulated. For example, aggregation levels for a Geography dimension might include City, County, State, Region, and so forth.

alert: In Oracle Demand Planning, a set of exception conditions that generates a notification to specified individuals when the conditions are met.

agreement: A contract with a customer that serves as the basis for work authorization. An agreement may represent a legally binding contract, such as a purchase order, or a verbal authorization. An agreement sets the terms of payment for invoices generated against the agreement, and affect whether there are limits to the amount of revenue you can accrue or bill against the agreement. An agreement can fund the work of one or more projects.An arrangement with a customer that sets business terms for sales orders in advance. Oracle Order Management lets you assign pricing, accounting, invoicing and payment terms to an agreement. You can assign discounts to agreements that are automatically applied. You can refer to an agreement when you enter an order for a particular customer, and have relevant default values automatically fill in the order using standard value rule sets. see customer family agreement, generic agreement.

agreement, contract, price list: The standard transactions can have a reference to a contract number. This code may be used as a key to find a document containing the item's price. The appropriate Oracle document can be used in the PO change process to determine the source for the item's price. Full use of this document within the PO Change transaction needs to be reviewed.

agreement type: A classification for agreements. Reference agreement types in defining discounts or automatic note rules, classify your agreements to control selection of agreements during order entry, and for reporting purposes.

ahead: Quantities were delivered in advance of the customer's anticipated delivery date, or an over shipment in quantities occurred. The supplier must control this situation in such a way that he will not manufacture or deliver these quantities again. See: Behind.

AIAG: Automotive Industry Action Group, an organization which publishes combined EDI implementation requirements for the major automotive industry manufacturers and suppliers.

alert: A specific condition defined in Oracle Alert that checks your database and performs actions based on the information it finds there.

alert action: In Oracle Quality, an electronic mail message, operating system script, SQL script, or concurrent program request that is invoked when specified action rule conditions are met.

alert input: A parameter that determines the exact definition of an alert condition. You can set the input to different values depending upon when and to whom you are sending the alert. For example, an alert testing for users to change their passwords uses the number of days between password changes as an input. Oracle Alert does not require inputs when you define an alert.

alert output: A value that changes based on the outcome at the time Oracle Alert checks the alert condition. Oracle Alert uses outputs in the message sent to the alert recipient, although you do not have to display all outputs in the alert message.

ALL_TAB_COLUMNS: A standard Oracle Database Table, maintained by the database, that contains the definition of every Column in every Table and View.

ALL_VIEWS: A standard Oracle Database Table, maintained by the database, that contains the definition of every View. Item Type is a high level grouping of Processes. Item Types are identified by name (e.g. ORDER) and are used to group other entities (like activities and messages) according to the type of business transaction or document they work with. Access to Item Types may be constrained by setting the R/W/X role privileges.

allocation: In Oracle Demand Planning, the process of changing child nodes values based on the change in the parent node, according to an algorithm. The algorithm might assign fixed, often equal, weights to each child, with the weights summing to unity.

agreement: A contract with a customer that serves as the basis for work authorization. An agreement may represent a legally binding contract, such as a purchase order, or a verbal authorization. An agreement sets the terms of payment for invoices generated against the agreement, and affect whether there are limits to the amount of revenue you can accrue or bill against the agreement. An agreement can fund the work of one or more projects.An arrangement with a customer that sets business terms for sales orders in advance. Oracle Order Management lets you assign pricing, accounting, invoicing and payment terms to an agreement. You can assign discounts to agreements that are automatically applied. You can refer to an agreement when you enter an order for a particular customer, and have relevant default values automatically fill in the order using standard value rule sets. see customer family agreement, generic agreement.

agreement, contract, price list: The standard transactions can have a reference to a contract number. This code may be used as a key to find a document containing the item's price. The appropriate Oracle document can be used in the PO change process to determine the source for the item's price. Full use of this document within the PO Change transaction needs to be reviewed.

agreement type: A classification for agreements. Reference agreement types in defining discounts or automatic note rules, classify your agreements to control selection of agreements during order entry, and for reporting purposes.

ahead: Quantities were delivered in advance of the customer's anticipated delivery date, or an over shipment in quantities occurred. The supplier must control this situation in such a way that he will not manufacture or deliver these quantities again. See: Behind.

AIAG: Automotive Industry Action Group, an organization which publishes combined EDI implementation requirements for the major automotive industry manufacturers and suppliers.

alert: A specific condition defined in Oracle Alert that checks your database and performs actions based on the information it finds there.

alert action: In Oracle Quality, an electronic mail message, operating system script, SQL script, or concurrent program request that is invoked when specified action rule conditions are met.

alert input: A parameter that determines the exact definition of an alert condition. You can set the input to different values depending upon when and to whom you are sending the alert. For example, an alert testing for users to change their passwords uses the number of days between password changes as an input. Oracle Alert does not require inputs when you define an alert.

alert output: A value that changes based on the outcome at the time Oracle Alert checks the alert condition. Oracle Alert uses outputs in the message sent to the alert recipient, although you do not have to display all outputs in the alert message.

ALL_TAB_COLUMNS: A standard Oracle Database Table, maintained by the database, that contains the definition of every Column in every Table and View.

ALL_VIEWS: A standard Oracle Database Table, maintained by the database, that contains the definition of every View. Item Type is a high level grouping of Processes. Item Types are identified by name (e.g. ORDER) and are used to group other entities (like activities and messages) according to the type of business transaction or document they work with. Access to Item Types may be constrained by setting the R/W/X role privileges.

allocation floor: See lowest allocation level.

alpha: In Oracle Demand Planning, a parameter used by the three forecasting methods of the exponential smoothing family: single, double, and Holt-Winters. This is the “level,” or baseline, parameter.

alpha smoothing factor: A value between 0 and 1 used in statistical forecasting calculations for smoothing demand fluctuations. Inventory uses the factor to determine how much weight to give to current demand when calculating a forecast.

anomalies: Outliers and other non-standard patterns in historical data.

APS: Acronym for Advanced Planning and Scheduling, a suite of Oracle products.

assembly: An item that has a bill of material. You can purchase or manufacture an assembly item. See also Bill of Material.

assignment: In Oracle Demand Planning, the slice of data (measures and other dimension values) that a demand planner is responsible for forecasting and submitting to the shared database. Can also include data that the planner can view but not submit. See also assignment scope, view scope.

assignment scope: In Oracle Demand Planning, the portion of the data (measures and other dimension values) that a demand planner is responsible for submitting to the shared database.

automatic method: See best fit forecast.

B

base model: The model item from which a configuration item was created.

best fit forecast: The best performing method of the statistical forecasting models, based on historical data.

beta: Parameter used by two forecasting methods of the exponential smoothing type: double and Holt-Winters. It is referred to as the “trend” parameter because it controls the estimate of the trend.

Bill of Material (BOM): A list of component items associated with a parent item and information about how each item relates to the parent item. Oracle Manufacturing supports standard, model, option class, and planning bills. The item information on a bill depends on the item type and bill type. The most common type of bill is a standard bill of material. A standard bill of material lists the components associated with a product or subassembly. It specifies the required quantity for each component plus other information to control work in process, material planning, and other Oracle Manufacturing functions. Also known as product structures.

bias: An error in the mean value of a forecast when compared to the actual data. Bias could apply to the entire data or to a segment of the data.

BOM item type: An item classification that determines the items you can use as components in a bill of material. BOM Item types include standard, model, option class, and planning items.

C

calendar: A system to handle consistency for time-varying data. Oracle Demand Planning usually uses a Manufacturing calendar, where production is divided into four week periods.

calibration: The process of estimating model parameters from historical data in statistical forecasting models. Alternatively known as training.

cannibalization: The phenomena, often observed during the introduction of new products, in which one product takes the market share of another related product.

causal analysis: The process of forecast generation that uses known or calculated relations between a dependent variable, such as demand, and one or more factors that affect the dependent variable, such as weather, product promotion, and discounts.

cell protection: In Oracle Demand Planning, locking a cell so that its data cannot be modified until the protection is removed.

child: In Oracle Demand Planning, a dimension value at the level immediately below a particular value in a hierarchy. Values of children are included in the calculation that produces the aggregated total for a parent. A dimension value might be a child for more than one parent if the dimension has more than one hierarchy. See also hierarchy.

collaborative forecasting: In Oracle Demand Planning, forecast generated by consultation among, or consolidation of, individual forecasts issued by organizations that are linked to each other in the supply chain such as suppliers, manufacturers, customers, retailers and manufacturers.

column: An Oracle Demand Planning report or worksheet has three components for displaying multi-dimensional data: column, row, and page. The column component separates data for the dimension values that run across the report. See also across.

competitive event: Event that could potentially compete with other events in terms of its effect on product demand. In Oracle Demand Planning, only the competitive event with the highest priority is applied in any particular area.

component demand: Demand passed down from a parent assembly to a component.

component item: An item associated with a parent item on a bill of material.

composite: In the Analytical Workspace used by Oracle Demand Planning, a dimension-like object that is derived from one or more base dimensions. Each member of the composite is a combination of members from each base dimension. Composite values are created automatically when data is assigned to a variable that uses the composite, thus reducing the number of NA values that would be stored on disk

confidence bounds or limits: The maximum and minimum level of variation expected for demand. These bounds are usually dictated by the uncertainty associated with the forecasts and represent the degree of confidence in the forecast values. For statistical forecasts, these correspond to the range with which a forecast value is expected to lie with some probability.

configuration: A product a customer orders by choosing a base model and a list of options. It can be shipped as individual pieces as a set (kit) or as an assembly (configuration item).

configuration bill of material: The bill of material for a configuration item.

configuration item: The item that corresponds to a base model and a specific list of options. Bills of Material creates a configuration item for assemble-to-order models.

configure-to-order: An environment where you enter customer orders by choosing a base model and then selecting options from a list of choices.

consensus forecast: Forecast generation by consultation among or consolidation of individual forecasts issued by different groups within an organization.

consolidation: The process of merging multiple forecasts for related, but not identical quantities. The forecasts can originate from multiple sources such as sales forecasts for different items, management forecasts for individual products, two managers forecasting sales for two different regions, or statistical forecasts for different commodities. Forecasts might also originate from a single source such as sales forecasts of different products from one salesperson.

constrained forecast: Unconstrained forecasts are generated from historical data or by individuals without regard to limitations. For example, a retailer with no knowledge of manufacturing capability or a manufacturer without knowledge might generate an unconstrained forecast. When an unconstrained forecast is appropriately modified based on knowledge of constraints, it is known as a constrained forecast. See also unconstrained forecast.

correlation: The relation between two or more variables, expressed as a number between -1 (perfect negative relation) and +1 (perfect positive correlation). The number 0 implies no relation.

correlation of events: See event correlation.

cross-correlation: The relation between a variable and the lagged values of another variable, expressed as a number between -1 (perfect negative relation) and +1 (perfect positive correlation). The number 0 implies no relation.

currency: The units of demand for that are used during data visualization, manipulation, and forecasting. In Oracle Demand Planning, currency is referred to as measures. See also measure.

custom aggregate: In Oracle Demand Planning, a single value that planners and the Demand Plan Manager can create to combine multiple values from one dimension into a single value. For example, a planner might create a custom aggregate that consists of three geographies rolled up into a single geography value. Custom aggregates appear on the user's selection list of dimension values.

cyclical decay: A parameter for linear and nonlinear regression that is useful if the length of the history is large and some cyclical component has been identified. The parameter value indicates how seriously deviations from baseline activity are considered: a higher value implies slower decay while a lower value implies faster decay for cyclical components. The allowable range is from 0.2 to 1.0. Since the algorithm searches for the best value of the parameter within the specified range, the recommendation is to leave the default values. Note that for less history (for example, less than about 1.5 to 2 years) and in the absence of cyclical activity, this parameter might not have any effect on the calculated forecasts.

D

data element: A variable or set of variables that store data values. Alternatively, a place holder for data storage or viewing such as a cell or set of highlighted cells.

decimal precision: Number of digits after the decimal point that will be displayed (with rounding).

demand plan: In Oracle Demand Planning, an overall forecast of demand and possibly plans for integration with a supply chain. This consists of all available information from the OLAPI Server including measures, baseline forecasts, conversion tables, scenarios, and event information.

dependent demand: Products for which forecasts are generated as a percentage or fractional rate of a related product. The rates might be obtained from history, from user input, or from the Planning Bill of Materials (BOM). For example, the blade of a turbine or the moon roof of a car illustrate dependent demand. See also independent demand, planning percent.

dependent variable: In regression or modeling, a variable that is expressed as a function of another variable, or explained in terms of another variable.

differencing: The process of obtaining difference values at any give time. Difference values are the current value less the value at a given lag.

dimension: In Oracle Demand Planning, a list of categories for data. A dimension acts as an index for identifying the values of a variable. Examples of dimensions are Product, Geography, and Time.

dimension value: In Oracle Demand Planning, an element in the list that makes up a dimension. For example, dimension values in the Geography dimension might include Boston, Chicago, and New York.

disaggregate: The inverse of aggregation; spreading down or allocating data values at higher nodes to the lower nodes.

distribution: In Oracle Demand Planning a term used to refer to the process by which specific slices data are allocated to planners, based on their assignments.

document: In Oracle Demand Planning a collective term used to refer to a report, graph, or worksheet. See also report, graph, worksheet.

down: The row position in a report or worksheet. Rows go down the page. See also row.

drill: In Demand Planning, to navigate up and down through the levels of aggregation in a dimension that has a hierarchy. When selecting dimension values or viewing data, you can expand or collapse a hierarchy by drilling down or up in it. Drilling down expands the view to include child values that are associated with parent values in the dimension hierarchy. Drilling up collapses the list of descendant values associated with a parent value in the dimension hierarchy.

E

error bars: Error range that forecast values are expected to exhibit with some degree of confidence.

estimation: Defining or determining a model or the parameters thereof.

event: In Oracle Demand Planning, any occurrence that is expected to impact the sales or demand forecast. For example, promotions, price reductions, product phase outs, and product introductions might be events.

event model: A technique that attempts to numerically capture the effect of an event on product demand at any aggregation level.

event correlation: The relation among events, such as a product introduction event and the corresponding cannibalization of a related product.

exception report: In Oracle Demand Planning, an ad hoc report that displays data associated with dimension values that are exceptions to planner-defined cutoff values. The criteria is applied to each page of data. For example, an exception report might identify customers whose orders have dropped by more a certain amount compared to the same period last year.

exponential smoothing: A class of statistic forecasting techniques in which the forecasts are obtained as simple linear weights of the actual value and the forecast value at previous time steps, or through some variation of this rule. The linear weights are the model parameters which are determined through calibration from historical data. Demand Planning offers three exponential smoothing methods: single, double, and triple (referred to as “Holt-Winter's”).

F

filtering: In statistics, smoothing the data values by removing high frequency components. Filtering aids in the identification of underlying longer term trends and seasonality in the data that have some forecast-ability.

final assembly order: A discrete job created from a configuration or an assemble to order item and linked to a sales order. Also known as final assembly schedule.

forecast error: Each forecast that Demand Planning generates includes an estimate of the forecast error. The types of errors computed are Mean Absolute Percent Error (MAPE), Mean Square Error (MSE), and Mean Absolute Deviation (MAD). See also Mean Absolute Percent Error (MAPE), Mean Square Error (MSE), Mean Absolute Deviation (MAD)

forecast reconciliation: See reconciliation.

forecasting methods: Refers to the statistical methods that you can use to generate forecasts in Oracle Demand Planning. These are Linear Regression, Polynomial Regression, Exponential Fit, Logarithmic Fit, Asymptotic fit, Exponential Asymptotic Fit, Single Exponential Smoothing, Double Exponential Smoothing, and Holt-Winters. As an alternative to selecting a specific method, you can allow the system to automatically determine the best-fit method.

formula measure: In Oracle Demand Planning, a formula that uses existing values in the database to calculate data on the fly. For example, you might create a formula measure that applies a weight to a forecast measure. A formula measure is not stored in the database, has no unit of measure association, and its values are never editable. See also measure.

frequency: The inverse of the time period at which the historical or forecast data are expected to repeat their past behavior. Seasonal or periodic data exhibit a low frequency component corresponding to their period. Purely random data that contain no substantial information usually exhibit high frequency.

frequency domain analysis: Analyzing time series data by first transforming it to the frequency domain, or by determining the time series components as a function of frequency. Estimates of the frequency components of a time series is often achieved through smoothed periodograms, using Fourier analysis. Frequency domain analysis often provides a better estimate for seasonal or periodic cyclicity in the data than time domain analysis. Demand Planning uses frequency domain analysis to calculate signal to noise ratios for outlier detection; however, this analysis is not visible to the planner and results are not reported directly.

G

gamma: A parameter used by Holt-Winters, a forecasting method in the exponential smoothing family. This is the seasonal parameter.

Geneva: Refers to Roadmap Geneva Forecasting™ from Roadmap Technologies. Geneva is the forecasting engine that is used in Oracle Demand Planning for statistical forecast generation.

geography dimension: In Oracle Demand Planning, a category of data that denotes geographical area. For example, the Geography dimension might include values such as “Eastern US,” “Massachusetts,” and “Boston.” See also dimension.

graph: An Oracle Demand Planning document that enables you to visualize and manipulate multi-dimensional data in graphical format. Types of graphs include area, bar, line, bar-line, pie, 3D, and scatter graphs.

H

hierarchy: In Oracle Demand Planning, a means of organizing and structuring data within a dimension. A hierarchy exists when values within a dimension are arranged in levels, with each level representing the aggregated total of the data from the below. For example, a Geography dimension might have a hierarchy that includes levels for Account, City, State, and Region. See also aggregation, level.

I

independence: See independent variable.

independent demand: Products for which demand forecasts are generated from their own data, which could be in the form of historical demand, causal factors, third party information, and so forth. For example, forecasts for finished goods such as cars and computers illustrate independent demand. See also independent demand.

independent variable: If a variable does not depend on another variable or set of variables in any way, then the first variable is said to be independent of the second. In regression, independent variable are those that cannot be expressed in terms of another variable. Note that independence implies no correlation between variables; however, the converse is not necessarily true.

input errors: Errors in measurement or data entry of variables that are used as inputs in the forecasting or planning process. Input error can be due to human or instrument error.

intermittent demand: Refers to sporadic demand patterns observed for certain products over certain time horizons. These types of demand are characterized by sudden surges followed by periods of zero or very low demand. Special modeling efforts might be required to handle these situations if the products constitute a significant proportion of the total sales.

introduction: See product introduction event.

item: Anything you make, purchase, or sell, including components, subassemblies, finished products, or supplies. Oracle Manufacturing also uses items to represent planning items that you can forecast, standard lines that you can include on invoices, and option classes you can use to group options in model and option class bills.

L

level: In Oracle Demand Planning, a position in a dimension hierarchy. Each level above the base level represents the aggregated total of the data from the level below. For example a Geography dimension might have ascending levels such as Customer, City, State, and Region. Within a dimension hierarchy, a dimension value at one level has a family relationship with the dimension values at the levels above and below that level. See also aggregation, hierarchy.

life cycle

The period over which a product stays in the market or is operationally produced by the manufacturer. A product's lifecycle events include the following:

Born — A new product
Young — A recently introduced product
Old — A mature product
Dead — A product whose functionality is no longer required as a result of new technology

likelihood: The probability of occurrence. The likelihood of a given model is the probability of the assumed model and the estimated parameters approximating reality.

linear models: A class of linear formulations that approximate the future behavior of a variable in terms of its own current and past values, or the current and past values of independent variables. The term implies that the model expresses the forecasts or the dependent variable as a linear function of the independent variables.

linear regression: A forecasting method in which a linear relationship () is fitted to the data, where a and be are parameters estimated from history.

lowest allocation level: In Oracle Demand Planning, a setting in the OLAPI Server that determines the lowest level of each dimension to which data will be allocated for a measure.

M

MAD: See Mean Absolute Deviation (MAD).

mandatory event: An event whose effects will be felt independently of other events that might occur. In Oracle Demand Planning, a priority is associated with such an event to determine the precedence of occurrence. This priority is necessary as the demand modification adds a specific number or uses an uplift factor, and operator precedence becomes important.

MAPE: See Mean Absolute Percent Error (MAPE).

make-to-order: An environment where customers order unique configurations that must be manufactured using multiple discrete jobs and/or final assembly orders where the product from one discrete job is required as a component on another discrete job. Oracle Manufacturing does not provide special support for this environment beyond the support it provides for assemble-to-order manufacturing.

mandatory component: A component in a bill that is not optional. Bills of Material distinguishes required components from options in model and option class bills of material. Mandatory components in pick-to-order model bills are often referred to as included items, especially if they are shippable.

Mean Absolute Deviation (MAD): A performance metric for evaluating forecast accuracy, where N is the total number of observations, the summation is for all N observations, Y represents the observed data at any time, and Z is the corresponding forecast data.

Mean Absolute Percent Error (MAPE): A performance metric for evaluating forecast accuracy. where N is the total number of observations, the summation is for all N observations, Y represents the observed data at any time, and Z is the corresponding forecast data.

Mean Squared Error (MSE): A performance metric for evaluating forecast accuracy. , where N is the total number of observations, the summation is for all N observations, Y represents the observed data at any time, and Z is the corresponding forecast data.

measure: In Oracle Demand Planning, a data variable that is a placeholder for storing or calculating data values for all levels of all dimensions. See also stored measure and formula measure.

measurement errors: The expected errors or standard deviation errors that are associated with the measured or input values. For example, errors during a transaction are measurement errors.

metric: See performance metric.

middle-out: A method of allocating forecasts generated at a middle level of a hierarchy to the higher levels through aggregation and to lower levels through allocation.

model: An abstract, often statistical or mathematical, representation of reality. Used to approximate the complex process of demand generation, forecast generation, and answer what-if queries.

model bill of material: A bill of material for a model item. A model bill lists option classes and options available when you place an order for the model item.

model item: An item whose bill of material lists options and option classes available when you place an order for the model item.

moving average: A smoothing or forecasting technique where the smoothed or forecast value at a point is determined by the average, or weighted average for weighted moving average, of adjacent values in time. The window size used for computing the average is often a parameter that must be specified. For smoothing, both lagged and lead values can be used; for forecast, only past values can be used.

moving median: A smoothing or forecast technique where the smoothed or forecast value at a point is determined by the median of adjacent values in time. The window size used for computing the median is often a parameter that must be specified. It is used in Demand Planning to detect and filter outliers from historical data during statistical forecast generation. However, this analysis is hidden from the planner and results are not directly reported.

MSE: See Mean Squared Error.

MRP: See material requirements planning.

multidimensional: In Oracle Demand Planning, refers to accessing data through multiple dimensions that can be arranged and organized according to a planner's requirements.

N

new product introduction: See product introduction event.

noise: In statistics, random variations in data caused by small changes in unaccountable factors, such as measurement error and lack of perfect repeatability during data generation. Loosely used to denote unexplained variances that are associated with demand that could be caused by individual human factors and are not expected to reoccur.

nonlinear regression: A class of statistical forecasting methods in which a linear relationship is fitted to a transformation of the original data. Demand Planning offers five non-linear regression methods: asymptotic fit, exponential fit, exponential asymptotic fit, logarithmic fit, and polynomial regression.

O

ODP: Abbreviation for Oracle Demand Planning.

OLAP: Abbreviation for Online Analytical Processing.

option: An optional item component in an option class or model bill of material.

option class: A group of related option items. An option class is orderable only within a model. An option class can also contain included items.

option class bill of material: A bill of material for an option class item that contains a list of related options.

option class item: An item whose bill of material contains a list of related options.

Oracle Workflow: The Oracle Workflow engine sequences the processes of demand planning. It is also used to schedule jobs and notify planners of certain events or exceptions.

outlier: A data value that is unusually large or small, usually caused by events that are not expected to reoccur and must be removed from the historical data when generating statistical forecasts.

outlier detection and filtering: The process of detecting and removing outliers from historical data for purposes of statistical forecast generation. The statistical forecasting techniques in Demand Planning automatically filter outliers through one of two methods: moving median in the “time domain” and signal to noise ratio in the “frequency domain.” During the forecast generation process, the best method is automatically selected.

P

page: In an Oracle Demand Planning document, the page component separates data for the dimension values that appear as pages.

parameter: Adjustable variable in a forecasting model, the values of which must be adjusted from historical data through a process called calibration or training.

parameter estimation: See calibration.

parent: In a hierarchical relationship in Oracle Demand Planning, the immediately superior level or node. For example in a Geography dimension, Eastern U.S. might be the parent of Massachusetts.

pareto analysis

A method for forecast model selection based on the type of item in stock. Items are divided into the following categories:

A — Comprising 1 to 20% of all items but representing 80% of the total value, forecast in a controlled environment with non-adaptive forecasting models and significant monitoring.
B — Comprising 20 to 30% of all items worth 15 to 20% of sales, forecast using short term adaptive models.
C — Comprising 50 to 70% of all items and 5 to 10% of the value for which forecasts cannot provide significant ROI.

percolation: The process of modifying the values of parent nodes and successive nodes at the top of a hierarchy to reflect changes in the values of child nodes.

performance metric: A quantity that measures the performance of forecast values by comparing with actual data, for example, root mean square error.

period type: The temporal granularity, such as weeks or months, at which the effects of an event are likely to be felt.

periodicity: See seasonality.

personal database: In Oracle Demand Planning, the database to which a demand planner has modification rights. A demand planner works in his or her personal database and submits final forecasts to the shared database.

phase out: See product phase out event.

planning bill of material: A bill of material for a planning item that contains a list of items and planning percentages. You can use a planning bill to facilitate master scheduling and/or material planning. The total output of a planning bill of material is not limited to 100% (it can exceed this number by any amount).

planning percent: A component usage percentage that facilitates planning for optional components on model and option class bills, and all components on planning bills.

point value: The value of a variable defined at any one point, such as at a given level of aggregation, for specified hierarchies of each dimension.

predefined reports: In Oracle Demand Planning, a set of preformatted reports that target specific issues. There are predefined reports for forecast accuracy, trend analysis, comparison, distribution, growth, quota, and ranking.

price list: A list containing the base selling price per unit for a group of items, item categories or service offered. All prices in a price list are for the same currency.

product dimension: In Oracle Demand Planning, a category of data that denotes a product or group of products. For example, you might have product dimension values such as “sporting goods” and “rackets.” See also dimension.

product introduction event: An event in which a new product is introduced. Oracle Demand Planning calculates the forecast based on the history or forecast, depending on the model type, of one or a linear combination of many model products using the spread model of a spread model product, which might be the same as the model product.

product phase out event: An event in which an existing product is allowed to die. In Oracle Demand Planning, a product phase out event is modeled as linear decay in demand.

promotion event: An event that represents the introduction of special schemes, incentives, or advertising to promote the demand and sale of a particular product. In Oracle Demand Planning, a promotion is defined as a standard event.

R

recalculate: In Oracle Demand Planning, refers to calculating forecast values at selected points based on modified or new planner inputs, while keeping other forecast parameters the same as before.

reconciliation: Forecasts issued at higher levels of aggregation must be propagated to the lower levels through a top down technique. However, this process must not interfere with the numbers at the lower level, which could be generated from another forecast or for which some actual data might be available. In Demand Planning, reconciliation refers to the process of propagating forecast information from higher to lower levels without loss of information at the lower levels.

reforecast: Generate new forecast values for a selected data range, using modified history, a new history date range, or a different forecast method. A reforecast might also take into account effects such as promotions and new product introductions that had not been accounted for in a previous forecast.

regression: In statistics, a technique for determining the mathematical relation and the associated uncertainties between dependent variables, such as demand; independent variables, such as past demand; and other factors, such as price, promotions and discounts, through linear or nonlinear models. The form of the model could be dictated beforehand, or it could be dictated by the data.

report: A tabular presentation of multidimensional data. Demand Planning supports ad hoc reporting and also includes a set of predefined reports.

residual: Remainder. The residual of a forecasting process refers to the forecast error or the remaining variability that cannot be explained by the forecasting process.

responsibility: Determines the data, windows, menus, reports, and concurrent programs you can access in Oracle Applications. It is linked directly to a data group. Several users can share the same responsibility, and a single user can have multiple responsibilities.

RMSE: See Root Mean Squared Error

ROI on forecast: Return of Investment on forecast. Compares the revenue saved by issuing accurate forecast for n units of an item with the cost of gathering the data and issuing the forecast.

role: See responsibility.

Root Mean Squared Error (RMSE): A number that summarizes the overall error and reflects the variability in the data. The individual errors are squared, added together, dividing by the number of individual errors, and then square rooted. A large RMSE indicates high fluctuations in differences between actual sales or shipments to corresponding forecast data.

row: An Oracle Demand Planning report has three components for displaying multidimensional data: column, row, and page. The row component separates data for the dimension values that run down the report. See also down.

S

safety stock: Amount of inventory set aside to cover shortages. Depends on the forecast uncertainties and past performance.

scenario: In Oracle Demand Planning, forecasts can be generated from multiple sources such as sales, management, marketing, budget, constrained, and unconstrained. Forecasts from each source can also differ in scope, certainty, expectation (for example, optimist versus pessimistic), aggregation level, time horizon, attributes, and so forth. Each unique situation or function represents a scenario. Scenarios are set up and associated with a demand plan in the OLAPI Server. Planners submit a forecast for each scenario in the demand plan.

seasonality: Repetitive pattern from year to year with demand for some periods considerably higher than others.

shared database: In Oracle Demand Planning, the database that stores consolidated data for a demand plan. Planners submit forecasts from their personal databases to the shared database; the Demand Plan Manager reviews data in the shared database before uploading the data to the OLAPI Server.

signal to noise ratio: In time series, the ratio of the strength of the underlying signal, which is thought to generate the data, and the noise, which is thought to corrupt the data. This is a frequency domain tool used in Demand Planning for outlier detection. See also noise.

sporadic demand: See intermittent demand.

spread model: The method for allocating data using top-down, bottom-up, or middle-out strategies. This could be done using specified weights, or weights that are calculated from history or forecasts.

standard event: In Oracle Demand Planning, any event other than product introduction or product phase out. These are characterized by event category, event type, priority, modification type, and uplift factor. The modifications are either additions or subtractions of numbers, or a factor of the demand of the product under consideration.

statistical forecast: Historical data are used to generate statistical forecasts in Oracle Demand Planning. The resultant forecasts include an estimate of the forecast error.

submit: In Oracle Demand Planning, refers to the action that a planner takes when he or she selects a forecast to commit to the shared database. Planners work on forecasts in their personal databases. Once a forecast is selected for submission, the data are moved from the personal database to a consolidated measure in the shared database.

supply chain planning: The development and maintenance of multi-organizational distribution and manufacturing plans across a global supply chain.

T

time dimension

In Oracle Demand Planning, a dimension whose values represent time periods. For example, values in the time dimension could include the following:

Years such as “2003” and “2004”
Quarters such as “Quarter 1 - 2003” and “Quarter 2 - 2003”
Months such as “February 2003” and “March 2003”

time domain analysis: Analysis of time dependent quantities using the actual values in time as independent variables. Temporal qualifiers such as time elapsed from some start date, might also be used as independent variables.

time series: A set of time-stamped data, such as demand data by the day.

time series analysis: Statistical analysis of time series data to determine statistical and other properties, and to often generate forecasts.

top-down: A method of allocating forecasts generated at the highest level of a hierarchy to the lower levels through allocation by a specified spreading rule.

tracking: The process of monitoring demand, historical data, accuracy of historical data, previous forecasts, and the past performance of forecasts.

trend: A characteristic that measures the inclination of time varying, demand, or other data values, to move up or down in an approximate straight line. While trend usually implies linear trend, it is loosely used to indicate other characteristics of data such as variation from mean, periodicity, or nonlinearity.

trend analysis: The process of analyzing data to yield trend estimates.

U

uncertainty

The degree of belief associated with a forecast. Uncertainty is useful to understand the expected skill of each forecast, and is an essential input for issuing composite forecasts. Uncertainty is measured and reported in terms of one of the following:

the expected standard deviation of the forecast errors
the expected upper and lower bonds on the forecasts
the probability of the forecast being accurate

unconstrained forecast: Forecast that is generated from historical data or by individuals without regard to limitations. For example, a retailer with no knowledge of manufacturing capability or a manufacturer without knowledge of customer demand might generate an unconstrained forecast. See also constrained forecast.

uplift factor: For a standard event, the factor by which the existing demand must be increased or decreased.

upload: In Oracle Demand Planning, refers to the action that a Demand Plan Manager takes when he or she commits a final forecast to the OLAPI Server.

V

variable: In Oracle Demand, a quantity that acts as a placeholder for a single datum or group of data. Demand for a particular product in a specific region is an example of a variable. See also measure.

variance: In statistics, the square of the standard deviation. In Demand Planning, variance is loosely used to denote the forecast errors.

variant: In Oracle Demand Planning, a forecast version and related what-if analyses that planners can create.

view: See multidimensional.

view scope: In Oracle Demand Planning, refers to the portion of the data that a demand planner can view.

verification: The process of determining forecast performance using data that was not used during the forecast process.

what-if analysis: A mechanism that enables planners to formulate ad hoc queries that represent hypothetical but realistic situations.

workflow: See Oracle Workflow.

worksheet: An Oracle Demand Planning document that enables planners to view and modify data.