• the first form is used by EViews native ( “EViews and MicroTSP”) and time series database formats ( “Time Series Formats”).

• the second form used for raw data files—Excel, Lotus, ASCII text, and binary files ( “Raw Data Formats”).

• the third form is used with the remaining source formats, which we term dataset formats, since the data have already been arranged in named variables ( “Datasets”).

• In most cases, the external data source is a file, so the source_description will be the description of the file (including local path or URL information, if necessary). Alternatively, the external data source may be the output from a web server, in which case the URL should be provided. Similarly, when reading from an ODBC query, the ODBC DSN (data source name) should be used as the source_description.

• Where there is more than one table that could be formed from the specified external data source, a table_description may be provided to select the desired table. For example, when reading from an Excel file, an optional cell range may be provided to specify which data are to be read from the spreadsheet. When reading from an ODBC data source, a SQL query or table name must be used to specify the table of data to be read.

• In raw data formats, the table_description allows you to provide additional information regarding names and descriptions for variables to be read, missing values codes, settings for automatic format, and data range subsetting.

• When working with text or binary files, the table_description must be used to describe how to break up the file into columns and rows.

• The @keep and @drop keywords, followed by a list of names and patterns, are used to specify variables to be retain or dropped. Similarly, the @keepmap and @dropmap keywords followed by lists of name patterns controls the reading of value labels. The keyword @selectif, followed by an if condition (e.g., “if age>30 and gender=1”) may be used to select a subset of the observations in the original data. By default, all variables, value labels, and observations are read.

• “range = arg”, where arg is a range of cells to read from the Excel workbook, following the standard Excel format [worksheet!][topleft_cell[:bottomright_cell]].

• “byrow”, transpose the incoming data. This option allows you to read files where the series are contained in rows (one row per series) rather than columns.

• “colhead=int”, number of table rows to be treated as column headers.

• “namepos = [first|firstatt|last|lastatt|all|none|attonly|discard|custom]”, which row(s) of the column headers should be used to form the column name, and also how to use the rest. The setting “first” (or “last”) refers to the object name being in the first (or last) column header row, and all other rows as the object's description. Similarly, “firstatt” (or “lastatt”) will use the first (or last) row as the name field, but will use all others as a custom attribute. The setting “all” will concatenate all column header fields into the object's name. “none” will concatenate all column header fields into the object's description. “attonly” will save all column header fields into the object's custom attributes. “discard” will skip all header rows altogether, and “custom” will allow you to specify explicitly how to treat each column header row using the “colheadnames=” argument. The default setting is “all” if no “colheadnames=” is specified, otherwise “custom”.

• "colheadnames = ("arg1", "arg2")", required when “namepos=custom”. Specifies the name & type of each column header row. “Name” will be mapped to the object name, “Description” to the object's description field, and the rest will be stored as custom object attributes. Any blank name will cause that column header row to be skipped.

• “nonames”, the file does not contain a column header (same as “colhead=0”).

• “names=("arg1","arg2",…)”, user specified column names, where arg1, arg2, … are names of the first series, the second series, etc. when names are provided, these override any names that would otherwise be formed from the column headers.

• “descriptions=("arg1","arg2",…)”, user specified descriptions of the series. If descriptions are provided, these override any descriptions that would otherwise be read from the data.

• “types=("arg1","arg2",…)”, user specified data types of the series. If types are provided they will override the types automatically detected by EViews. You may use any of the following format keywords: “a” (character data), “f” (numeric data), “d” (dates), or “w” (EViews automatic detection). Note that the types appear without quotes: e.g., “types=(a,a,a)”.

• “na="arg1"”, text used to represent observations that are missing from the file. The text should be enclosed on double quotes.

• “scan=[int| all]”, number of rows of the table to scan during automatic format detection (“scan=all” scans the entire file). Note: If a “range=” argument is not specified, then EViews will only scan the first five rows of data to try and determine the data format for each column. Likewise, if the “na=” argument is not specified, EViews will also try to determine possible NA values by looking for repeated values in the same rows. If the first five rows are not enough to correctly determine the data format, use the “scan=” argument to instruct EViews to look at more rows. In addition, you may want to specify a the “na=” value to override any dynamic NA value that EViews may determine on its own.

• “firstobs=int”, first observation to be imported from the table of data (default is 1). This option may be used to start reading rows from partway through the table.

• “lastobs=int”, last observation to be read from the table of data (default is last observation of the file). This option may be used to read only part of the file, which may be useful for testing.

• “table = arg”, where arg specifies which table to read in an HTML file/page containing multiple tables.

• “skip = int”, where int is the number of rows to discard from the top of the HTML table.

• “byrow”, transpose the incoming data. This option allows you to import files where the series are contained in rows (one row per series) rather than columns.

• “colhead=int”, number of table rows to be treated as column headers.

• “namepos = [first|firstatt|last|lastatt|all|none|attonly|discard|custom]”, which row(s) of the column headers should be used to form the column name, and also how to use the rest. The setting “first” (or “last”) refers to the object name being in the first (or last) column header row, and all other rows as the object's description. Similarly, “firstatt” (or “lastatt”) will use the first (or last) row as the name field, but will use all others as a custom attribute. The setting “all” will concatenate all column header fields into the object's name. “none” will concatenate all column header fields into the object's description. “attonly” will save all column header fields into the object's custom attributes. “discard” will skip all header rows altogether, and “custom” will allow you to specify explicitly how to treat each column header row using the “colheadnames=” argument. The default setting is “all” if no “colheadnames=” is specified, otherwise “custom”.

• "colheadnames = ("arg1", "arg2")", required when “namepos=custom”. Specifies the name & type of each column header row. “Name” will be mapped to the object name, “Description” to the object's description field, and the rest will be stored as custom object attributes. Any blank name will cause that column header row to be skipped.

• “nonames”, the file does not contain a column header (same as “colhead=0”).

• “names=("arg1","arg2",…)”, user specified column names, where arg1, arg2, … are names of the first series, the second series, etc. when names are provided, these override any names that would otherwise be formed from the column headers.

• “descriptions=("arg1","arg2",…)”, user specified descriptions of the series. If descriptions are provided, these override any descriptions that would otherwise be read from the data.

• “types=("arg1","arg2",…)”, user specified data types of the series. If types are provided they will override the types automatically detected by EViews. You may use any of the following format keywords: “a” (character data), “f” (numeric data), “d” (dates), or “w”(EViews automatic detection). Note that the types appear without quotes: e.g., “types=(a,a,a)”.

• “na="arg1"”, text used to represent observations that are missing from the file. The text should be enclosed on double quotes.

• “scan=[int|all]”, number of rows of the table to scan during automatic format detection (“scan=all” scans the entire file). Note: If a "range=" argument is not specified, then EViews will only scan the first five rows of data to try and determine the data format for each column. Likewise, if the "na=" argument is not specified, EViews will also try to determine possible NA values by looking for repeated values in the same rows. If the first five rows are not enough to correctly determine the data format, use the "scan=" argument to instruct EViews to look at more rows. In addition, you may want to specify a the "na=" value to override any dynamic NA value that EViews may determine on its own.

• “firstobs=int”, first observation to be imported from the table of data (default is 1). This option may be used to start reading rows from partway through the table.

• “lastobs = int”, last observation to be read from the table of data (default is last observation of the file). This option may be used to read only part of the file, which may be useful for testing.

• “ftype = [ascii|binary]” specifies whether numbers and dates in the file are stored in a human readable text (ASCII), or machine readable (Binary) form.

• “rectype = [crlf|fixed|streamed]” describes the record structure of the file:

• “reclines =int”, number of lines to use in forming each row when “rectype=crlf” (default is 1).

• “reclen=int”, number of bytes to use in forming each row when “rectype=fixed”.

• “recfields=int”, number of fields to use in forming each row when “rectype=streamed”.

• “skip=int”, number of lines (if rectype is “crlf”) or bytes (if rectype is not “crlf”) to discard from the top of the file.

• “comment=string“, where string is a double-quoted string, specifies one or more characters to treat as a comment indicator. When a comment indicator is found, everything on the line to the right of where the comment indicator starts is ignored.

• “emptylines=[keep|drop]”, specifies whether empty lines should be ignored (“drop”), or treated as valid lines (“keep”) containing missing values. The default is to ignore empty lines.

• “tabwidth=int”, specifies the number of characters between tab stops when tabs are being replaced by spaces (default=8). Note that tabs are automatically replaced by spaces whenever they are not being treated as a field delimiter.

• “fieldtype=[delim|fixed|streamed|undivided]”, specifies the structure of fields within a record:

• “quotes=[single|double|both|none]”, specifies the character used for quoting fields, where “single” is the apostrophe, “double” is the double quote character, and “both” means that either single or double quotes are allowed (default is “both”). Characters contained within quotes are never treated as delimiters.

• “singlequote“, same as “quotes = single”.

• “delim=[comma|tab|space|dblspace|white|dblwhite]”, specifies the character(s) to treat as a delimiter. “White” means that either a tab or a space is a valid delimiter. You may also use the abbreviation “d=” in place of “delim=”.

• “custom="arg1"”, specifies custom delimiter characters in the double quoted string. Use the character “t” for tab, “s” for space and “a” for any character.

• “mult=[on|off]”, to treat multiple delimiters as one. Default value is “on” if “delim” is “space”, “dblspace”, “white”, or “dblwhite”, and “off” otherwise.

• “endian = [big|little]”, selects the endianness of numeric fields contained in binary files.

• “string = [nullterm|nullpad|spacepad]”, specifies how strings are stored in binary files. If “nullterm”, strings shorter than the field width are terminated with a single zero character. If “nullpad”, strings shorter than the field width are followed by extra zero characters up to the field width. If “spacepad”, strings shorter than the field width are followed by extra space characters up to the field width.

• “byrow”, transpose the incoming data. This option allows you to import files where the series are contained in rows (one row per series) rather than columns.

• “lastcol”, include implied last column. For lines that end with a delimiter, this option adds an additional column. When importing a CSV file, lines which have the delimiter as the last character (for example: “name, description, date”), EViews normally determines the line to have 3 columns. With the above option, EViews will determine the line to have 4 columns. Note this is not the same as a line containing “name, description, date”. In this case, EViews will always determine the line to have 3 columns regardless if the option is set.

• Fortran Format:

• Column Range Format:

• C printf/scanf Format:

• “colhead=int”, number of table rows to be treated as column headers.

• “namepos = [first|firstatt|last|lastatt|all|none|attonly|discard|custom]”, which row(s) of the column headers should be used to form the column name, and also how to use the rest. The setting “first” (or “last”) refers to the object name being in the first (or last) column header row, and all other rows as the object's description. Similarly, “firstatt” (or “lastatt”) will use the first (or last) row as the name field, but will use all others as a custom attribute. The setting “all” will concatenate all column header fields into the object's name. “none” will concatenate all column header fields into the object's description. “attonly” will save all column header fields into the object's custom attributes. “discard” will skip all header rows altogether, and “custom” will allow you to specify explicitly how to treat each column header row using the “colheadnames=” argument. The default setting is “all” if no “colheadnames=” is specified, otherwise “custom”.

• "colheadnames = ("arg1", "arg2")", required when “namepos=custom”. Specifies the name & type of each column header row. “Name” will be mapped to the object name, “Description” to the object's description field, and the rest will be stored as custom object attributes. Any blank name will cause that column header row to be skipped.

• “nonames”, the file does not contain a column header (same as “colhead=0”).

• “names=("arg1", "arg2",…)”, user specified column names, where arg1, arg2, … are names of the first series, the second series, etc. when names are provided, these override any names that would otherwise be formed from the column headers.

• “descriptions=("arg1", "arg2",…)”, user specified descriptions of the series. If descriptions are provided, these override any descriptions that would otherwise be read from the data.

• “types=("arg1","arg2",…)”, user specified data types of the series. If types are provided they will override the types automatically detected by EViews. You may use any of the following format keywords: “a” (character data), “f” (numeric data), “d” (dates), or “w” (EViews automatic detection). Note that the types appear without quotes: e.g., “types=(a,a,a)”.

• “na="arg1"”, text used to represent observations that are missing from the file. The text should be enclosed on double quotes.

• “scan=[int|all]”, number of rows of the table to scan during automatic format detection (“scan=all” scans the entire file). Note: If a “range=” argument is not specified, then EViews will only scan the first five rows of data to try and determine the data format for each column. Likewise, if the “na=” argument is not specified, EViews will also try to determine possible NA values by looking for repeated values in the same rows. If the first five rows are not enough to correctly determine the data format, use the “scan=” argument to instruct EViews to look at more rows. In addition, you may want to specify a the “na=” value to override any dynamic NA value that EViews may determine on its own.

• “firstobs=int”, first observation to be imported from the table of data (default is 1). This option may be used to start reading rows from partway through the table.

• “lastobs = int”, last observation to be read from the table of data (default is last observation of the file). This option may be used to read only part of the file, which may be useful for testing.