topic Re: Interval match and syntethic table in QlikView

Interval match and syntethic table

Tue, 21 Jul 2015 06:36:35 GMT

Hallo,

I hope that you guys can help me or at least give a hint.

I was about to create the 'bridge table' with the help of IntervalMatch:

events:

LOAD * INLINE [

event, eventdate

a, 12.03.2014

b, 12.03.2015

];

Interval:

LOAD * INLINE [

intervalId, begin, end, preis

1, 01.01.2013, 01.05.2014, 65

2, 01.01.2013, 01.05.2014, 55

3, 02.05.2014, 31.12.2015, 75

];

bridge:

IntervalMatch (eventdate)

load begin, end

Resident Interval;

Nevertheless I constantly get the synthetic table.

I have found other discussion where the similar example is shown but without synth table...

here is the data model:

Thank You in advance!

Re: Interval match and syntethic table

Anonymous — Tue, 21 Jul 2015 06:40:58 GMT

Hello,

Synthetic table is recommended in interval match scenarios. At least I recommend through my experience Best possible performance because if you join it into the bridge table (which you can) it multiplies the table and creates a huge one, slowing the performance eventually. As long as the result is correct, this model is best practice.

You can try joining it; you can see yourself. Just write "join" before the bridge table.

Serhan

Re: Interval match and syntethic table

jfkinspari — Tue, 21 Jul 2015 06:46:25 GMT

Include a left join before intervalmatch to get rid of the synthetic dimension.

bridge:

LEFT JOIN (Interval)

IntervalMatch (eventdate)

load begin, end

Resident Interval;

Re: Interval match and syntethic table

datanibbler — Tue, 21 Jul 2015 06:47:00 GMT

Hi Ira,

IntervalMatch is a two-step process in fact:

- First you use the function to join the intervals (start_date and end_date) to the distinct event_dates;

- Then you can use those two fields (plus a key_field, optionally) to join in the price_data from the intervals_table.

=> The synthetic key after the first step is normal (you haven't yet joined) (think about it, the tables have two fields which are named the
same and QlikView automatically links those, so of course you'd get a synthetic key)

Serhan is completely right. Write a JOIN (or I would recommend adding the name of the table you want that to be joined to in brackets like >> JOIN (events) << before the IntervalMatch() to join that to the events_table and the synthetic key will be gone.

=> Then you can, in a second step, join the price_data from the intervals_table to that.

HTH

Re: Interval match and syntethic table

Anonymous — Tue, 21 Jul 2015 07:01:05 GMT

Hi Gurus,

Not only that, I claim it would work better with syn table

Please note that and tell me your own experience I claim syn table is life-saver in many real-life scenarios with proper data.

Serhan

Re: Interval match and syntethic table

jfkinspari — Tue, 21 Jul 2015 07:10:37 GMT

I can understand your argument.

I have not made my own experience with huge interval tables, where it could affect performance.

In this case the keyfield is data, so it will probably newer get huge. In this case a clean data model might be preferable

Re: Interval match and syntethic table

Tue, 21 Jul 2015 11:09:22 GMT

Thank you for your answers But I have one more question... How can one avoid synthetic table, when slowly changing dimension is a case? In the example above appears synthetic table...

Thank you in advance

events:

LOAD * INLINE [

event, eventdate, intervalId

a, 12.03.2014, 1

b, 12.03.2015, 3

];

Interval:

LOAD * INLINE [

intervalId, begin, end, preis

1, 01.01.2013, 01.05.2014, 65

2, 01.01.2016, 01.05.2016, 55

3, 02.05.2014, 31.12.2015, 75

4, 02.05.2014, 31.12.2015, 85

];

bridge:

right join

IntervalMatch (eventdate, intervalId)

load begin, end, intervalId

Resident Interval;