Set Analysis performance: Are numeric filters faster than string based filters?

[mwallman]

Hi,

In the old versions of Qlik I read somewhere that numeric values are faster to process than string based.

Is this the case for Set Analysis too?

Let's say you have an expression like SUM({<CarStatus={"Sold"}>}CarSales)

Is it better to replace the Sold with 1 or some value associated with the string?

The above example is quite simple but if you imagine a complex app where there are multiple filters in set analysis, and you want to improve the performance I was wondering if something like this is better.

Would the performance of Set Analysis be better doing a change like this?

[sunny_talwar]

Based on testing in QlikView (10,000,000 row of data), it seems that there might be a very very small advantage of using number vs text.

Table:

LOAD *,

If(CarStatus = 'Sold', 1, 0) as CarStatusNum;

LOAD RecNo() as Key,

Ceil(Rand() * 100) * 100 as CarSales,

If(Rand() < 0.5, 'Sold', 'Not Sold') as CarStatus

AutoGenerate 10000000;