How to convert a numeric list into a character one?

[#1]

Hi everybody,

I have a list containing numbers and I would like to transform them into characters. Is there a way to do that without using a loop? I was not able to find the proper function.

Thanks for your help!

[#2]

Here is a little script that creates a function that does what you want.  You can either define and use the function in your code, or use the 4 statements in open code to do the conversion.

Names Default To Here( 1 );
theList = {7,8,9};

NumListToChar = Function( {myList},
	{defaultlocal},
	dtx = New Table( "x", New Column( "y", values( myList ) ), private );
	dtx:y << data type( character );
	myList = dtx:y << get values;
	Close( dtx, nosave );
	Return( myList );
);

theList = NumListToChar( theList );

[#3]

Thanks for your answer @txnelson it is indeed what I was looking for.

I just wonder what is the faster between a loop and a function which involves table creation. Any idea?

[#4 · Accepted solution]

New in JMP 16 is the For Each() function which iterates over a container – list, matrix, or associative array – to perform an operation on each element.

Here is an example to show how to use it to convert your list of numbers to a list of character strings.

Names Default To Here( 1 );
theList = {7, 8, 9};

For Each( {value, index}, theList, theList[index] = Char( value ) );

[#5]

Thanks @Jeff_Perkinson !

I did some simulations just by curiosity to compare the execution time of each solution and it seems that the for each function is on average the fastest one!

So I am gonna pick that one although the function of @txnelson also works very well.

Thanks to both of you

[#6]

Here's a quick timing program for the three methods, for loop, for each, and custom function using data table.

Names Default To Here( 1 );

listsize=1000000;

theList = As List( J( listsize, 1, Random Integer( 100 ) ) );

NumListToChar = Function( {myList},
	{defaultlocal},
	dtx = New Table( "x", New Column( "y", values( myList ) ), private );
	dtx:y << data type( character );
	myList = dtx:y << get values;
	Close( dtx, nosave );
	Return( myList );
);

start = HP Time();
For Each( {value, index}, theList, theList[index] = Char( value ) );
foreachloop = HP Time() - start;

theList = As List( J( listsize, 1, Random Integer( 100 ) ) );

start = HP Time();
theList = NumListToChar( theList );
customfunction = HP Time() - start;

theList = As List( J( listsize, 1, Random Integer( 100 ) ) );

start = HP Time();
For( i = 1, i <= N Items( theList ), i++,
	theList[i] = Char( theList[i] )
);
forloop = HP Time() - start;

show(foreachloop, customfunction, forloop);

For me with the listsize shown above the results are very similar between the three methods:

foreachloop = 1106974;
customfunction = 1077862;
forloop = 1400643;

[#7]

Extending Jeff's answer

Different overheads hurt for small sizes and help for large sizes.

dt = New Table( "Untitled 3",
	Add Rows( 0 ),
	New Column( "size", Numeric, "Continuous", Format( "Best", 12 ), Set Values( [] ) ),
	New Column( "for", Numeric, "Continuous", Format( "Best", 12 ), Set Values( [] ) ),
	New Column( "foreach", Numeric, "Continuous", Format( "Best", 12 ), Set Values( [] ) ),
	New Column( "custom", Numeric, "Continuous", Format( "Best", 12 ), Set Values( [] ) )
);

NumListToChar = Function( {myList},
	{defaultlocal},
	dtx = New Table( "x", New Column( "y", values( myList ) ), private );
	dtx:y << data type( character );
	myList = dtx:y << get values;
	Close( dtx, nosave );
	Return( myList );
);

For( listsize = 1, listsize < 1e7, listsize *= 2, 

	theList = As List( J( listsize, 1, Random Integer( 100 ) ) );

	start = HP Time();
	For Each( {value, index}, theList, theList[index] = Char( value ) );
	foreachloop = HP Time() - start;

	theList = As List( J( listsize, 1, Random Integer( 100 ) ) );

	start = HP Time();
	theList = NumListToChar( theList );
	customfunction = HP Time() - start;

	theList = As List( J( listsize, 1, Random Integer( 100 ) ) );

	start = HP Time();
	For( i = 1, i <= N Items( theList ), i++,
		theList[i] = Char( theList[i] )
	);
	forloop = HP Time() - start;

	Show( foreachloop, customfunction, forloop );
	dt << addrows( 1 );
	dt:size[N Rows( dt )] = listsize;
	dt:for[N Rows( dt )] = forloop / 1e6;
	dt:foreach[N Rows( dt )] = foreachloop / 1e6;
	dt:custom[N Rows( dt )] = customfunction / 1e6;
	Wait( 0.1 ); // so it shows on display
);

[#8]

Thanks for the additional information @Jeff_Perkinson and @Craige_Hales !

We always learn a lot of things with the Community!!

[#9]

Nice illustration of the size-dependent performance of the different functions. I'm still using JMP 15 and can't use For Each() yet, but to extend Jeff's timing exercise further I toss in two more functions that give the same result. 

Repeat(k = 0; {k++; Char( theList[k] )}, listsize);

Words( Substr( Char( theList ), 2, Length( Char( thelist ) ) - 2 ), ", " );

The For loop, which is the fastest for small lists, is surpassed by Words at size>50, by the data-table-detour function at size>1000 and by Repeat at size>30000.

I was a little surprised to see that Words remained the fastest for size>50, about twice as fast at millions of "words".

[#10]

Wow I was looking for one solution and I got so many! JMP is amazing! Thanks all of you!