Conditional Functions

JSL provides five functions to evaluate an expression conditionally: If(), Match(), Choose(), Interpolate(), and Step().

The If() function evaluates the first result expression when its condition evaluates as true (a nonzero or nonmissing value). Otherwise, it evaluates the second result expression.

The syntax is:

If (condition, result1, result2)

For example, the following script returns "Young" when the age is less than 12. Otherwise, the script returns "Young at Heart".

You can also string together multiple conditions and results. The syntax is:

If (condition1, result1,

condition2, result2,

...,

resultElse);

In the preceding example, if condition1 is not true, the function continues evaluating until it finds a true condition. Then that condition’s result is returned.

The last result is returned when all conditions are false. And when a value is missing, the missing value is returned. For these reasons, it’s very important to include a default result at the end of the expression. Consider the following example, which recodes gender abbreviations in Big Class.jmp:

For Each Row(sex =

If(

sex == "F", "Female",

sex == "M", "Male",

"Unknown");

);

The script works like this:


For Each Row(sex =	For each row in the table, sex is the column that is recoded.
If(	Begins the If() loop.
sex == "F", "Female",	If the value of sex is F, replaces the value with Female.
sex == "M", "Male",	If the value of sex is M, replaces the value with Male.
"Unknown");	If neither of the above conditions are true, replaces the value with Unknown. If this result were omitted and the value of sex were missing, the script would return a missing value.
);	Ends the loop.

You can also put actions and assignments in the result expression. The following example assigns 20 to x, because the first condition (y < 20) is false:

y = 25;

z = If( y < 20, x = y, x = 20 );

Note: Be careful to use two equal signs (==) for equality tests, not one equal sign (=). An If with an argument such as name=value assigns rather than tests the value.

Match

You can use the Match() function to make several equality comparisons without needing to rewrite the value to be compared. The syntax is:

Match(x, value1, result1, value2, result2, ..., resultElse)

For example, the following script recodes gender abbreviations in Big Class.jmp:

The script works like this:


For Each Row(sex =	For each row in the table, sex is the column that is recoded.
Match(	Begins the Match() loop.
sex,	Specifies sex as the match argument.
"F", "Female",	If the value matches "F" replace it with “Female”.
"M", "Male",	If the value matches "M", replace it with "Male.
"Unknown");	If F or M are not matched, replaces the value with "Unknown".
);	Ends the loop.

This Match() example is a simplified version of the example in If. The advantage of Match() is that you define the comparison value once rather than repeat it in each condition. The disadvantage is that you cannot use expressions with operators as you can with If; the argument sex == "F" returns an error in a Match() expression.

With more groups of conditions and results, the value of Match() becomes more apparent. The following script would require many additional lines of code with If().

dt=open("$SAMPLE_DATA/Travel Costs.jmp");

For Each Row(Booking Day of Week =

Match(Booking Day of Week, "Sunday", "SUN", "Monday", "MON", "Tuesday", "TUE", "Wednesday", "WED", "Thursday", "THU", "Friday", "FRI", "Saturday", "SAT", "Not Specified");

);

Be careful with the data type of the condition and result. In the preceding example, the conditions and results are both character data. If the data types do not match, JMP automatically changes the column’s data type. The results are not what you want.

The following script changes the column’s data type from numeric to character based on the first cell’s data type. The first value, "12", is replaced with "Twelve", and the remaining cells are filled with "Other".

dt=open("$SAMPLE_DATA/Big Class.jmp");

For Each Row(age =

Match(age, 12, "Twelve", 13, "Thirteen", 14, "Fourteen", 15, "Fifteen", 16, "Sixteen", "Other");

);

When data consists of integers such as 1, 2, and 3, you can save even more typing by using Choose(). See Choose for more information.

Choose

The Choose() function shortens scripts even more than Match(), provided the arguments are tested against integers. The syntax is:

Choose(expr, result1, result2, result3, ..., resultElse)

Suppose you have a data table with a column of numeric values from 1 through 7. If the first cell contains the number 1, the following script returns x = "Low".

The script works like this:


x =	Creates the x variable.
Choose(	Begins the Choose() loop.
group,	Evaluates the value of group.
"Low",	If the value of group is 1, return "Low".
"Medium",	If the value of group is 2, return "Medium".
"High",	If the value of group is 3, return "High".
"Unknown"	Otherwise, return "Unknown".
)	Ends the loop.
);	Closes the x variable.
Show(x);	Returns the value of x.

If the expression evaluates to an out-of-range integer (such as 7 when only 4 replacement values are listed), the last result is returned. In the preceding example, "Unknown" is returned.

Notice that If() and Match() require more code to achieve the same results as the Choose function:

if(group==1, "Low", group==2, "Medium", group==3, "High", "Unknown");

match(group, 1, "Low", 2, "Medium", 3, "High", "Unknown");

Note: If the data types in the expression do not match, JMP automatically changes the column’s data type.

Interpolate

The Interpolate() function finds the y value corresponding to a given x value between two points (x1, y1 and x2, y2). A linear interpolation is applied to the values. You might use Interpolate() to calculate missing values between data points.

The data points can be specified as a list:

Interpolate(x, x1, y1, x2, y2, ...)

or as matrices containing the x and y values:

Interpolate(x, xmatrix, ymatrix)

Suppose that your data set includes the height of individuals from age 20 through 25. However, there is no data for age 23. To estimate the height for 23-year-olds, use interpolation. The following example shows the value that you want to evaluate (age 23), followed by matrices for ages (20 through 25) and heights (59 through 75).

Interpolate(23, [20 21 22 24 25], [59 62 56 69 75]);

returns:

62.5

The value 62.5 is halfway between the y values 56 and 69, just as 23 is halfway between the x values 22 and 24.

Notes:

•	The data points in each list or matrix must create a positive slope. For example, Interpolate(2,1,1,3,3) returns 2. However, Interpolate(2,3,3,1,1) returns a missing value (.).

•	Interpolate is best used for continuous data, but Step() is for discrete data. See Step for details.

Step

The Step() is like Interpolate() except that it finds the corresponding y for a given x from a step-function fit rather than a linear fit. Use Step() with discrete y values (that is, when the y value’s corresponding x value can be only y1 or y2). However, when the y value’s corresponding x value can fall between y1 and y2, use Interpolate().

As with Interpolate, the data points can be specified as a list:

Step(x, x1, y1, x2, y2, ...)

or as matrices containing the x and y values:

Step(x, xmatrix, ymatrix)

Suppose that your data table shows the discount percentage for purchases of $25, $50, $75, and $100. You want to create a graph that shows the discount for a $35 purchase, which the data table does not specify. The following example shows the value that you want to evaluate, 35, followed by matrices for purchases from $25 to $100.

Step(35, [25 50 75 100], [5 10 15 25]);

returns:

If the discounts were on a sliding scale (in this example, between 5 and 10, you would use Interpolate():

Interpolate(35, [25 50 75 100], [5 10 15 25]);

returns:

As with Interpolate(), the data points must create a positive slope.