Data Source Options

To display this dialog, double-click on a data source node .

A data source node is the starting point of each stream; it loads data from a single file into the stream's database.

 

Text file

Loads the database from a CSV formatted file.

If you select this option, you must also enter the data file and select a categorical field option.

 

Database via OLE DB

Loads the database from an OLE DB source.

 

All records

Loads all of the records available from the database.

 

Sample first n records

Samples the records whilst the file is being loaded. This method is more efficient than loading the whole database (which might not be possible due to memory limitations) and then sampling the records. Enter the number of records (n) that will be sampled from the database into the edit box.

 

Data file

The file from which the database will be loaded. Click on the select button to the right of the field to display the file selection dialog.

The file should be in comma delimited text format. There is no special representation for missing values - simply enter consecutive commas with no spaces between them.

 

Categorical fields

You can choose the way in which WITNESS Miner is alerted when it reads in a categorical value (that is, a text value) from the text file.

Select the use (*) flag in datafile option if you have prefixed each categorical (text) field heading in the CSV file with an asterisk. WITNESS Miner will be alerted to each categorical field as it loads the data.

Choose the select manually option if categorical fields are not marked with an asterisk in the CSV file. Then click on the select button to display the select categorical fields dialog, which enables you to select categorical field headings from a list of available field headings.

 

Connection String

If you are loading data from an OLE DB source, enter the database connection string, which specifies the data source, provider, security and so on. Click on the build button to invoke the data links wizard, which helps you to build the connection string.

 

SQL Query String / Table Name

If you are loading data from an OLE DB source, enter an SQL query string or table name that WITNESS Miner will use to build the record set. If an entity name contains spaces, you must enclose the name in square brackets.

For example, this SQL query string selects the Software and Response fields from the Customer Response file to build the database:

SELECT Software, Response FROM [Customer Response]

For example, this table name uses all the fields in the Customer Response file to build the database:

[Customer Response]

 

Load dates as continuous numeric values

If you are loading data from an OLE DB source, check this box to load date fields as continuous numeric values (doubles) rather than text strings. This means that you can sort the database easily by date fields. By default, this option is switched off.

If you are loading data from an OLE DB source and dates are stored as text, WITNESS Miner loads dates in the format that is specified by the currently-selected input locale. You can change this by using the Regional Settings options in the Windows Control Panel.

If you are loading data from an OLE DB source and dates are stored as real numbers, WITNESS Miner takes the number to the left of the decimal point as the number of days since January 1st 1900. The fractional part indicates the number of milliseconds since midnight.