Skip to main content

Read CSV

Synopsis​

This Operator reads an ExampleSet from the specified CSV file.

Description​

CSV is an abbreviation for Comma-Separated Values. The CSV files store data (both numerical and text) in plain-text form. All values corresponding to an Example are stored as one line in the CSV file. Values for different Attributes are separated by a separator character. The separator remains constant. Each row in the file uses the constant separator for separating Attribute values. The term 'CSV' suggests that the Attribute values would be separated by commas, but other separators can also be used.

The easiest way to import a CSV file is to use the Import_Configuration_Wizard from the Parameters panel. All parameters can also directly be set in the Parameters panel. For more details about the Operator, see the description of the parameters.

Please make sure that the CSV file is read correctly as an ExampleSet before building a Process that uses it.

Differentiation​

There are many Read <source> Operators in the Data Access group and Files/Read sub-group. For example, there is Read Excel, Read URL, Read SPSS, Read XML and other Operators, which can read ExampleSet from different file formats.

Input​

file​

A CSV file can be optionally passed in as a file object. This can be created with Operators having file output ports such as the Open File (Deprecated) Operator.

Output​

output​

This port delivers the ExampleSet created from the CSV file provided at the input port, imported through the Import_Configuration_Wizard or loaded from the path given to the csv_file parameter.

Parameters​

Import Configuration Wizard​

This user-friendly wizard guides you to easily configure this Operator to import the CSV file.

column separators​

column separators for CSV files can be specified here. It can also be provided as a regular expression. A good understanding of regular expressions can be developed by studying the description of Select Attributes operator and its tutorial Processes.

trim lines​

This parameter indicates if lines should be trimmed (removal of empty spaces at the beginning and the end) before the column split is performed. This option might be problematic if TABs ('\t') are used as separators.

multiline text​

Indicates if text values may span multiple lines. If checked, line breaks enclosed by quotes are treated as part of the data.

use quotes​

This parameter indicates if quotes should be regarded. Quotes can be used to store special characters like column_separators. For example if (,) is set as column_separators and (") is set as quotes_character, then a row (a,b,c,d) will be translated as 4 values for 4 columns. On the other hand ("a,b,c,d") will be translated as a single column value a,b,c,d. If this parameter is set to false, the quotes_character parameter and the escape_character parameter cannot be defined.

quotes character​

This parameter defines the quotes character and is only available if use_quotes is set to true.

escape character​

This parameter specifies the character used to escape the quotes and is only available if use_quotes is set to true. For example, if (") is used as quotes_character and ('') is used as escape character, then ("yes") will be translated as (yes) and ("yes") will be translated as ("yes").

skip comments​

This parameter is used to ignore comments in the CSV file (if any). If this option is set to true, a comment character should be defined using the comment_characters parameter.

comment characters​

This parameter is available if comment characters is set to true. Lines beginning with these characters are ignored. If this character is present in the middle of the line, anything that comes in that line after this character is ignored. The comment character itself is also ignored.

starting row​

This parameter defines which row is the first to be read (everything before will be skipped). This index is one based and includes the header row into the count if the option is set.

parse numbers​

This parameter specifies whether numbers are parsed or not.

decimal character​

This character is used as the decimal character.

grouped digits​

This parameter decides whether grouped digits should be parsed or not. If this parameter is set to true, a grouping_character parameter has to be specified.

grouping character​

This character is used as the grouping character. If this character is found between numbers, the numbers are combined and this character is ignored. For example if "22-14" is present in the CSV file and "-" is set as the grouping character, then "2214" will be stored.

infinity representation​

This parameter can be set to parse a specific infinity representation (e.g. "Infinity"). If it is not set, the local specific infinity representation will be used.

date format​

The parameter specifies the date and time format. Many predefined options exist but users can also specify a new format. If text in a CSV file column matches this date format, that column is automatically converted to date type.

Some corrections are automatically made on invalid date values. For example, a value '32-March' will automatically be converted to '1-April'.

Columns containing values which cannot be interpreted as numbers will be interpreted as nominal, as long as they do not match the date and time pattern of the date format parameter. If they match, this column of the CSV file will be automatically parsed as date and the corresponding Attribute will be of type date.

use header row​

If this parameter is set to true, the header_row parameter needs to specify the row with the names of the Attributes. If this is activated, no annotations can be used.

header row​

Sets the index of the row that should be used for Attribute names. The Attributes names are determined by the content of this row. As a consequence, data will only be read below this row. The parameter data_set_meta_data_information overwrites the data from the header row if it is set.

annotations​

If the use_header_row is not set to true, annotations can be added using the 'Edit List' button of this parameter, which opens a new menu. This menu allows you to select any row and assign an annotation to it. Name, Comment and Unit annotations can be assigned. If row 0 is assigned a Name annotation, it is equivalent to setting the first row as names parameter to true. If you want to ignore any row, you can annotate them as Comment. Remember that row number in this menu does not count commented lines.

time zone​

Users can select any time zone from the list of provided time zones. If text in the CSV file column matches the format provided in the date_format parameter, that column is automatically converted to date type (see date_format parameter info).

The converted date is interpreted to be provided in the time zone specified in this parameter.

locale​

Users can select any locale from the list of provided locales.

encoding​

Users can select any encoding from the list of provided encodings.

read all values as polynominal​

This option allows you to disable the type handling for this operator. Every column will be read as a polynominal attribute.

data set meta data information​

This parameter allows to adjust or override the meta data of the CSV file. Column index, name, type and role can be specified here.

The Read CSV Operator automatically tries to determine an appropriate data type of the Attributes by reading the first few lines and checking the occurring values. Integer values are assigned the integer data type, real values the real data type. Values which cannot be interpreted as numbers are assigned the nominal data type, as long as they do not match the format of the date_format parameter.

With the data set meta data information parameter, this automatic assignment can be adjusted or overwritten. This parameter also overwrites the information from the header_row.

read not matching values as missings​

If this parameter is set to true, values that do not match with the expected value type are considered as missing values and are replaced by '?'. For example, if 'abc' is written in an integer column, it will be treated as a missing value. A question mark (?) in the CSV file is also read as a missing value.