Conversions

Selecting this option opens a dialog when a factor column has a large number of missing values and is being read from a foreign data source such as Excel. The dialog lets you carry out a fill down action to copy values into cells with missing values.

Selecting this option opens a dialog when a column has repeated values, but has fewer unique values than the number specified here. The dialog lets you convert the columns to factors.

When you enter a numerical value into cell in a factor column, if the number is not an existing factor level, it will be assigned to the closest existing level, provided the difference between the entered number and that level is less than the provided tolerance (which must be positive). A value of 0 will always raise an error on any mismatch.

For example, you have existing factor levels 10, 20 and 30 and a tolerance of 1. Entering 9 or 11 is within tolerance of level 10, so Genstat will convert your value to this level. Any other values close to 10 will result in the error message below.

A positive value in this field will be used as a tolerance in creating levels from a factor. If you enter two values that differ by less than this tolerance, then they will be assigned to the same factor group.

The default value of 0 places all unique values into their own group. If the variate being converted to a factor contains calculated values, then these values may differ in their final few significant figures.

For example, the calculated values 1/3 and 200/600 may be assigned to different groups depending on how the final figure rounds. A small value such as 1e-12 could then be used to allocate these two distinct values to the same group.

If selected, when updating Genstat, any factor levels or labels that are not used will be removed. For example, if a factor column Weekdays does not contain an entry for Tuesday, the label Tuesday will be removed from the set of labels when you next update Genstat.

This option is only applied once, either when a column is new or is imported from a foreign source.

This opens the User Label Sort Order dialog, which allows a list of labels to be entered. If the labels in a factor match the user defined labels, then they will be sorted into the order in this list. Multiple orders may be entered by providing these in sections.

For example a colour order and a species order can be entered by providing the list of labels: Black, Blue, Green, Red, Yellow, White, followed by Clover, Lotus, Ryegrass, Cocksfoot.

Sometimes numbers are accidentally mistyped as letters. Common mistypings include i, I, l, L for the number 1; o, O for the number 0; a comma for a decimal point, etc. Text to number enables Genstat to convert text labels by substituting numbers for letters or, where this is not possible, the label is converted to a missing value.

• Strict – Change text strings to numbers only when there are no non-numeric characters in the string. For example, the text ’23’ is converted to 23, the text ‘1a’ contains an alphabetic character so the whole label is converted to a missing value.
• 1 substitution – Allow a single number to be mistyped and substitute this with the appropriate number. A single lower case ‘o’ or upper case ‘O’ is converted to 0: lower case ‘i’, upper case ‘I’, lower case ‘l’, or upper case ‘L’ is converted to 1. Lower case ‘s’ or upper case ‘S’ is converted to 2. A comma is converted to a decimal point. For example, ‘1o’ contains a single alphabetic character so is converted to 10, ‘1,i’ contains two a comma and an alphabetic characters so is converted to a missing value.
• Common substitution – Allows multiple numbers to be mistyped and converts them to the appropriate numbers. For example, ‘LOO’ is converted to 100.
• Standard – The same as Common substitution, but any letters at the end of a number are ignored. For example, ’23o’ is converted to 23, ‘i20’ has an alphabetic character at the front so the whole label is converted to a missing value.
• Lax – Change all text numbers to actual numbers and remove any non-numeric characters. For example, ‘A23X’ is converted to 23.

When reading an Excel xlsx file, this option controls what happens to Unicode characters that are not supported by the extended ASCII character set. Some Unicode characters, such as Greek or mathematical symbols, can be displayed using the Genstat typesetting commands used by the PRINT directive. However, many foreign language Unicode characters cannot be displayed by Genstat and these will be removed with a warning. The options below control how the supported Unicode characters are handled.

• Convert Unicode to Genstat typesetting – The Unicode characters are converted to equivalent typesetting strings, e.g. α is converted to ~{alpha} and √ to ~{sqrt}.
• Convert Unicode to ASCII characters – The Unicode characters are converted to the nearest ASCII characters, e.g. α is converted to a and √ to sqrt.
• Remove all Unicode characters – All Unicode characters are removed.

The conversion of Greek letters to ASCII characters is given in the table below.

When a variate created from a calculation is converted to a factor, re-calculating the column may generate some invalid factor levels. For example, if the column C1 contains the values 1, 2, 3 and 4 and the column C2 is calculated as C2 = 2*C1, this will produce the values 2, 4, 6 and 8. After converting C2 to a factor, if a value is changed in C1 to 5 this will create a recalculated value of 10 which is not a valid factor level (2, 4, 6 or 8) of C2. This item controls what happens when converting the variate to a factor:

• Always remove calculation – The column’s calculation will always be removed.
• Always keep calculation – The column’s calculation will always be kept.
• Prompt to remove or keep – A dialog will display, giving you a choice to remove or keep the column’s calculation