Terminology applicable to .
NOTE: This list is not comprehensive.
These terms apply to the objects that you import, create, and generate in .
A data type refers to the expected class of values for a column of data. A data type defines the types of information that are expected and can include specific formatting of that information. Column values that do not meet the expectations of the column data type are determined to be invalid for the data type.
A container for holding a set of related imported datasets, recipes, and output objects. Flows are managed in Flow View page.
A reference to an object that contains data to be wrangled in . An imported dataset is created when you specify the file(s) or table(s) that you wish to read through a connection.
A job is the sequence of processing steps that apply each step of your recipe in sequence across the entire dataset to generate the desired set of results.
A macro is a sequence of one or more reusable recipe steps. Macros can be configured to accept parameterized inputs, so that their functionality can be tailored to the recipe in which they are referenced.
Associated with a recipe, an output is a user-defined set of files or tables, formats, and locations where results are written after a job run on the recipe has completed.
An output may contain one or more destinations, each of which defines a file type, filename, and location where the results of the output are written.
A sequence of steps that transforms one or more datasets into a desired output. Recipes are built in the Transformer page using a sample of the dataset or datasets. When a job is executed, the steps of the recipe are applied in the listed order to the imported dataset or datasets to generate the output.
A pointer to the output of a recipe. A reference can be used in other flows, so that those flows get the latest version of the output from the referenced recipe.
A reference that has been imported into another flow.
A set of generated files or tables containing the results of processing a selected recipe, its datasets, and all upstream dependencies. See Job Details Page.
Optionally, you can create a profile of your generated results. This profile is available through the and may assist in analyzing or troubleshooting issues with your dataset. See Overview of Visual Profiling.
When you review and interact with your data in the data grid, you are seeing the current state of your recipe applied to a sample of the dataset. If the entire dataset is smaller than the defined limit, you are interacting with the entire dataset.
You can create new samples using one of several supported sampling techniques. See Overview of Sampling.
A schema defines the column names, data types, and ordering of your dataset. Schemas apply to relational datasources and some file types, such as Avro or Parquet. For more information, see Overview of Schema Management.
A set of columns, their order, and their formats to which you are attempting to wrangle your dataset. A target represents the schema to which you are attempting to wrangle. You can assign a target to your recipe, and the schema can be superimposed on the columns in the data grid, allowing you to make simple selections to transform your dataset to match the column names, order, and formats of the target. See Overview of Target Schema Mapping.
These terms apply to the , a web-based application for interacting with your datasets, flows, and recipes.
The main panel of Flow View where you can add, arrange, and remove flow objects. See Flow View Page.
Standardize values in a column by clustering similar values together. See Overview of Cluster Clean.
Creates a new column of data by providing example values from an existing column. See Overview of TBE.
Review sampled data across multiple columns through the column browser. You can also use the Column Browser panel to toggle the display of individual columns. See Column Browser Panel.
Examine details and profile of the data of a selected column. See Column Details Panel.
Perform transformation operations on the selected column from a list of menu options, including changing the column data type. See Column Menus.
At the top of the column, review the counts of values in the column. Select one or more values in the column through the histogram. See Column Histograms.
In the Transformer page, the data grid displays a sample of the dataset at the currently selected step in the recipe. Make selections in the dataset to prompt suggestions for transformations to add to your recipe. See Data Grid Panel.
Review color-coded counts of valid, missing, and mismatched values in your column based on the column's data type. Select a color bar to be prompted with suggestions for transformations on the relevant rows. See Data Quality Bars.
Change the data type for the column from the icon to the left of the column header. See Column Menus.
Examine details about your dataset, including source of data and other information. See Dataset Details Page.
The Flag for review feature enables flow users to flag recipe steps for others to review, provide inputs, and sign off on the changes before jobs are permitted to execute. See Flag for Review.
Create, manage, and export your flows. See Flows Page.
Build your flow objects, including recipes, outputs, and references. See Flow View Page.
Landing page after login. See Home Page.
Import data from a valid connection as an imported dataset. See Import Data Page.
Manage your imported datasets and reference objects. See Library Page.
Review the details of your job, including an optional profile of the resulting data. See Job Details Page.
Review the list of jobs that you have launched. View status, explore job details, and export results. See Job History Page.
Add, edit, and remove steps from your current recipe. Apply changes and see updates immediately in the data grid sample.
Configure job, visual profiling, and job outputs before launching. See Run Job Page.
Review, create, and delete samples for the current recipe.
Review status of all samples that you have initiated. Administrators can access the samples of all users. For more information, see Sample Jobs Page.
Feature that enables automated execution of flows according to user-defined schedules. See Overview of Scheduling.
Search for transformations to build as the next step in your recipe. See Search Panel.
Review and modify settings. See Preferences Page.
Standardize similar column values using multiple matching techniques in a simple interface. See Standardize Page.
Based on selections you make in the data grid, you can review profiling information and a set of suggested transformations to add to your recipe. See Selection Details Panel.
Feature that enables matching of columns and data types of your dataset with a pre-defined target schema.
Select from common transformations in a toolbar across the top of the data grid. See Transformer Toolbar.
Review and customize transformation steps. See Transform Builder.
Review sampled data, explore suggestions and previews, and build transformation steps. See Transformer Page.
Review and modify settings applicable to your user account. See User Profile Page.
Review and toggle the visibility of the columns in your dataset. See Visible Columns Panel.
As you build more complex recipes and flows, it's a good idea to create samples periodically in your recipe steps. All steps between the currently displayed sample and the currently displayed recipe step are executed in the browser, so this type of checkpointing with samples can improve performance. For more information on best practices in sampling, see Overview of Sampling.
The system within the for managing data types.
These terms pertain to building recipes in in the Transformer page.
An input to a function. See Wrangle Language.
Several functions can be used to group values in a column into bins, which can assist in preparing your data for downstream use. See Prepare Data for Machine Processing.
A data type is the set of constraints on expected values in a column. When you specify the data type for a column, you provide a means for the platform to identify the values in the column that do not match the selected type, which assists in wrangling the mismatched values. See Supported Data Types.
Data types can be selected from the column menus. See Column Menus.
An input to a recipe that is not the primary datasource for the recipe. For example, if your recipe includes a join step, the dataset that is joined into your recipe is an upstream dependency. Recipe steps and changes outside of the can create dependency errors, in which an upstream object can no longer be found and the reference to it cannot be resolved. These issues must be fixed prior to successful execution of a job. For more information, see Fix Dependency Issues.
A file's encoding defines the set of characters that are in use in the file. There are many different encoding systems in use around the world. To represent English language, which uses a 26-character alphabet, UTF-8 is sufficient. However, to represent Asian character sets, which may contain thousands of characters, a different and broader set of characters is required. See Supported File Encoding Types.
When a file is imported, assumes that the file is in the default encoding type. As needed, you can change the encoding type that is used to import the file. See Change File Encoding.
A full scan sample is generated across the entire dataset on default running environment. Full scan samples are more representative of the total dataset. However, they can take a while to generate. For more information, see Samples Panel.
A function in is an action that is applied to a set of values as part of a transformation step. A function can take 0 or more parameters as inputs, yielding a single output of a specific data type. For a list of supported functions, see Language Index.
When a file-based dataset is imported, attempts to detect the format and structure of the data and then to apply a set of initial parsing steps to transform the data for display in tabular form in the data grid. These steps may vary depending on the file format. See Initial Parsing Steps.
These steps do not appear in the recipe. As needed, you can disable the detection of structure on import. When disabled, these steps are added as the first steps of the recipe, where you can edit or remove them as needed. See Remove Initial Structure.
This database concept can be applied to datasets. In a join, two datasets are merged into one, based on a set of key columns. Values in these columns that match across the datasets are used to determined the values from each dataset to include in the joined dataset. See Join Types.
Joins are created as steps in your recipe. See Join Window.
A retrieval of a row of values from another dataset based on common values in columns in each dataset. A lookup is useful for bringing in reference information based on values in one of the columns of your dataset. See Lookup Wizard.
Values in a column that do not conform to range or format of expected values for the column's data type.
Cell values in the dataset that are empty.
A multi-dataset (MDS) operation refers to any step in your recipe that uses two or more datasets. Joins and unions are examples of multi-dataset operations.
An expression that is inside another expression. Example:
supports the use of nested expressions in your recipe steps. See Wrangle Language.
A value that does not exist in the dataset. See Manage Null Values.
A single character that represents an arithmetic function or comparison. For example, the Plus sign (
+) represents the add function.
|Logical Operators||and, or, and not operators|
|Numeric Operators||Add, subtract, multiply, and divide|
|Comparison Operators||Compare two values with greater than, equals, not equals, and less than operators|
|Ternary Operators||Use ternary operators to create if/then/else logic in your transforms.|
In statistics, an outlier refers to a value that is unusually above or below from the mean. In , an outlier is 4 standard deviations away from the mean.
You can review outliers for column values. See Column Statistics Reference.
An input to a transform in . See Wrangle Language.
In , a pattern is an object that describes a sub-string within a value. Patterns can be described using regular expressions, a common standard, or , a proprietary simplification of regular expressions. See Text Matching.
Patterns are widely used in the product for identifying and extract values from data, data type validation, and supporting pattern-based suggestions.
A quick scan sample is generated using an appropriate selection of rows from the dataset. Since these samples are generated in , they are faster to produce. For more information, see Samples Panel.
A range join is a type of join in which key values may be matched with a range of values in the joined-in dataset. For example, you can create a range join based on the source key value being greater than values in the key column of the joined-in dataset. A range join can explode the size of your resulting dataset. For more information, see Configure Range Join.
Joins are created as steps in your recipe. See Join Window.
Regular expressions are a powerful yet complex method of describing patterns of values for matching purposes. See Text Matching.
The row number for a record as it appeared in the original dataset. Source row number information can be obtained by function. This function may return a null value if multi-dataset operations, such as union and join, have been performed on the dataset. See SOURCEROWNUMBER Function.
A source metadata reference is a programmatic reference to some aspect of the source file for your dataset. Using these programmatic references, you can write source information for your original datasource into your dataset for future reference. For more information, see Source Metadata References.
provides multiple mechanisms to standardize column values using patterns, clustering algorithms, or functions. See Overview of Standardization.
String collation refers to a method of comparison of strings based on a set of rules. includes the following functions to perform string collation-based comparisons:
A transformation is the unit of action in a recipe step. A transformation applies one or more actions on a set of rows or columns. Transformations are specified in the Transformer page through the Transform Builder. See Transform Builder.
For a list of available transformations, see Transformation Reference.
A transform in is an action that is applied to rows or columns of your dataset. A transform can take zero or more parameters as inputs. A parameter may contain a reference to a column, a literal value, or a function.
NOTE: Transforms are not available through the . Instead, you build transformations, which are more complex steps that reference transforms from the underlying language.
For a list of supported transforms, see Language Index.
A simplification of regular expressions, are custom selectors for patterns in your data and provide a simpler and more readable alternative to regular expressions. See Text Matching.
A union combines two or more datasets such that the rows of the second and later datasets are appended to the end of the first dataset. In a union operation, the columns must be matched up, or the results are a ragged dataset.
Unions are created as steps in your recipe. See Union Page.
An informal term for the process of data preparation. Data wrangling was invented by the co-founders of .
Unix time (a.k.a. POSIX time or Epoch time) is a system for describing instants in time, defined as the number of seconds that have elapsed since 00:00:00 Coordinated Universal Time (UTC), Thursday, 1 January 1970, not counting leap seconds.