Nest transformation explicitly types transformed column
In prior releases, when a Nest transformation was applied to a column to nest values into Arrays or Objects, the resulting column was re-inferred by the Dataprep by Trifacta application. This re-inference should not be necessary, since the target column's data type is effectively declared in the transformation definition.
Beginning in this release, the output column of these Nest transformations is explicitly typed to Array or Object data type, based on the definition of the transformation.
NOTE: Existing uses of the Nest transformation are not immediately affected. However, if these transformations are edited, then the changes may cause unexpected results and breakages in downstream transformations. If the recipe was originally designed expecting a different data type, subsequent steps may have been used to clean up the nested data, assuming that it was String values or some other data type. If the output column is now explicitly typed as Array or Object data type, these steps may be broken. You may be able to fix these broken steps by explicitly typing the output column to String after the Nest transformation and before your subsequent steps.
For more information, see Nest Your Data.
Split by position no longer requires sorted list of positions
Beginning in this release, when you create a Split by position transformation, the numeric values indicating the positions do not need to be listed in sorted numeric order.
Tip: You can now do faster iteration since you can add new positions as needed when previewing the transformation.
For more information, see Split Column.
Support for numeric separators in NUMFORMAT function
Beginning in Release 8.5, the NUMFORMAT function supports the following configurable separators for localizing numeric values:
|Decimal Separator||The string used to separate the integer part of a Decimal value from its fractional part.|
|Grouping Separator||The string used to separate a group of digits.|
For more information, see NUMFORMAT Function.
Converts a string formatted as a number into an Integer or Decimal value by parsing out the specified decimal and group separators. A string or a function returning formatted numbers of String type or a column containing formatted numbers of string type can be inputs.
Documentation for the following functions is now available.
Returns the position of the nth occurrence of a letter or pattern in the input string where a specified matching string is located in the provided column. You can search either from left or right.
Evaluates an input against the String datatype. If the input matches, the function outputs a String value. Input can be a literal, a column of values, or a function returning values. Values can be of any data type.
Evaluates a String input against the Array datatype. If the input matches, the function outputs an Array value. Input can be a literal, a column of values, or a function returning String values.
Evaluates a String input against the Object datatype. If the input matches, the function outputs an Object value. Input can be a literal, a column of values, or a function returning String values.
Returns the serial date number for the last day of the month before or after a specified number of months from a starting date.
Transform Builder now supports All columns option
Beginning in Release 7.9, select All columns option has been added in the Transform Builder.
Selects all columns in the dataset
The following is the list of the transformations that accept the All option for selecting columns:
Replace text or patterns
Replace text between delimiters
Replace by position
Replace mismatched values
Replace missing values
Edit with formula
Change column type
For more information, see Transform Builder.
Rename transform now supports Upper / Lower and Left / Right options
Beginning in Release 7.8, the Rename transform supports the following new options:
|Convert to lowercase|
Converts existing column names to lowercase
|Convert to UPPERCASE||Converts existing column names to uppercase|
|Keep from beginning (left)||Specifies the number of characters to keep from the beginning of column names|
|Keep from end (right)||Specifies the number of characters to keep from the end of column names|
For more information on rename columns, see Rename Columns.
Approximation statistical functions:
Tip: Approximation functions are suitable for larger datasets. As the number of rows increases, accuracy and calculation performance improves for these functions.
Computes the approximate median from all row values in a column or group. Input column can be of Integer or Decimal.
Computes an approximation for a specified percentile across all row values in a column or group. Input column can be of Integer or Decimal.
Computes an approximation for a specified quartile across all row values in a column or group. Input column can be of Integer or Decimal.
base64 encoding functions:
Converts an input value to base64 encoding with optional padding with an equals sign (
Converts an input base64 value to text. Output type is String.
Derives the full name from a Datetime value of the corresponding weekday as a String. Source value can be a reference to a column containing Datetime values or a literal.
This page has no comments.