Generates for each row of a given input table a bit vector. The bit vectors are either generated from multiple numerical or string columns, a string column containing the bit positions to set, hexadecimal or binary strings or a collection column. In order to adjust the node settings please select first the source column object e.g. if the bit vector should be created from multiple string/numerical columns or from a single string/collection column. Depending on the selected option the corresponding dialog elements are enabled.

Bit vectors from a single column

In the case of a single input column only the selected single column to be parsed is considered for the generation of the bit vectors.

Single string column

In the case of a string input only the column containing the string is considered for the generation of the bit vectors. The string is parsed and converted into a bit vector. There are three valid input formats which can be parsed and converted:

• Hexadecimal strings: Strings consisting only of the characters 0-9 and A - F (where lower- or uppercase is not important). The represented hexadecimal number is converted into a binary number which is represented by the resulting bit vector.
• Binary strings: Strings consisting only of 0s and 1s are parsed and converted into the according bit vectors.
• ID strings: Strings consisting of numbers (separated by spaces) where the numbers refer to those positions in the bit vector which should be set. (Typical input format for association rule mining).

Single collection column

In the case of a single collection column each unique collection element gets a bit position assigned. The length of the bit vectors corresponds to the number of unique elements in a collection cells. For example if the input table contains two rows with the collections {a,b} and {b,c} the corresponding bit vectors will be [110] and [011].

Bit vectors from multiple columns

In the case of multiple columns the bit positions in the resulting bit vector correspond to the column position in the input table. For example if the second and third column of a given input table is selected and the first column is omitted the bit vectors of each row will have length 2. The first bit of the bit vector is set if the value of the second column matches the selected criterion likewise the second bit of the bit vector is set if the value of the third column matches the selected criterion. The columns to consider when creating the bit vector can be specified in the multiple column selection section. Using the enforce exclusion/inclusion option the node can be configured to handle previously unknown columns. If the enforce exclusion option is selected all unknown columns are added automatically to the include list whereas if the enforce inclusion option is selected all unknown columns are added to the exclude list. The columns to include can be also defined by a pattern if the Wildcard/Regex Selection option is selected.

Multiple string columns

The bit of a vector is set if the corresponding column value does match/does not match the specified pattern depending on the "Set bit if pattern does match/does not match" option. The pattern may contain wildcards such as '?' or '*' to match any one character or any sequence (including none) of characters. It can also be a complex regular expression.

Multiple numeric columns

There are two options to determine if the bit is set for the value in the corresponding column or not:

• Either a global threshold is defined, then all values which are above or equal to the threshold are converted into set bits, all other bit positions remain 0, or
• A certain percentage of the mean of each column is used as a threshold, then all values which are above or equal to the percentage of the mean are converted into set bits. As an example let's say the mean percentage is set to 50% and the mean of col1 is 2 and the mean of col2 is 8. Then the corresponding bit for col1 is set if the value is above or equal to 1 and for col2 if the value is above or equal to 4.

Missing values

For numeric data the incoming missing values will result in 0s. For multiple string columns a missing values will also result in 0s. For the string input missing values will also result in a missing value in the output table. If a string could not be parsed it will also result in a missing cell in the output table and an error message with detailed information is printed in the console. For a collection column all missing collection elements are ignored.

Options

Input column(s)

Select the input column(s) for creating bit vectors.

• Single string column: Create bit vectors from a single string column
• Single collection column: Create bit vectors from a single collection column
• Multiple string columns: Create bit vectors from multiple string columns
• Multiple numeric columns: Create bit vectors from multiple numeric columns

Pattern

The pattern to search for in the data value of each selected string column.

Case sensitive match

A case sensitive matching is performed if this option is selected.

Pattern matching

Select the pattern matching method.

• Manual: Define the pattern manually for simple string matching.
• Wildcard: Use wild cards in the pattern. Wildcard patterns contain '*' (matching any sequence of characters) and '?' (matching any one character).
• Regex: Specify a regular expression. Examples: "^foo.*" matches anything that starts with "foo". "[0-9]*" matches any string of digits. For a complete explanation of regular expressions, see the java.util.regex.Pattern class documentation.

Set bit if pattern

Sets the bit if the pattern:

• Matches: Set the bit for matching columns
• Mismatches: Set the bit for non-matching columns

Threshold

Choose how to define the threshold for setting bits when using multiple numeric columns.

• Global: Specify the global threshold. All values which are above or equal to this threshold will result in a 1 in the bit vector.
• Mean: Specify a percentage of the mean of each column which is used as threshold.

Global threshold

Specify the global threshold. All values which are above or equal to this threshold will result in a 1 in the bit vector.

Percentage of the mean

Specify which percentage of the mean a value should have in order to be set. For example, with a mean percentage of 50% and a column mean of 2, the bit is set if the value is above or equal to 1.

Column selection

Select the columns to include for bit vector creation. When creating bit vectors from multiple string columns, include string columns. When creating from multiple numeric columns, include numeric columns. The position of a column in the selection corresponds to the bit position in the resulting bit vector.

String representation format

Select one of the three valid input formats for the single string column:

• Bit string: A string of '0' and '1'
• Hexadecimal: A hexadecimal representation of bits
• IDs: A string of ids, where every id indicates the position in the bit set to set.

Column

The column to parse for creating bit vectors. Select a string column when using a single string column source, or a collection column when using a single collection column source.

Remove column(s) used for bit vector creation

If checked, the generating column(s) (the included columns for multi-column modes, or the selected single column) are removed from the output table. If unchecked, the generated bit vectors are appended to the input table.

Output column

The name of the output column containing the generated bit vectors.

Fail on invalid input

If selected, the node will fail during execution if a data cell could not be converted to a bit set. If unselected, the node will skip invalid entries and insert a missing value instead.

Bit vector type

The storage type for the generated bit vectors.

• Dense: Standard option recommended for dense vectors e.g. with more than 10% set bits
• Sparse: Option recommended for sparse vectors e.g. less than 10% set bits

Views

Statistics View

Provides information about the generation of the bit vectors from the data. In particular this is the number of processed rows, the total number of generated zeros and ones and the resulting ratio of 1s to 0s.

Workflows

• 00_Topic_Modeling_with_Verified_ComponentsKNIME Hub
• 00_Topic_Modeling_with_Verified_ComponentsKNIME Hub
• 01_ClusteringKNIME Hub
• 02_Topic_Scoring_ExampleKNIME Hub
• 02_Topic_Scoring_ExampleKNIME Hub
• Show all 90 workflows

Links

• No links available

Developers