Rolling Percentage Calculation

[Anonymous]

Hello,

I am trying to calculate a projected traffic number using the percentage listed on the table. For example, ABC.com's actual traffic starting from Dec-2017 is 300 and we expect a 25% increase in Jan-2018. Therefore, ABC.com's projected traffic would be 375. Then we take 375 and apply the 10% increase for Feb-2018 projected traffic number, and repeating this step for subsequent months. Is this possible to replicate in the script in Qlikview?

This table example is the desired output in Qlikview.

domain	month_year	percentage	actual traffic	projected traffic
ABC.com	Dec-2017	-	300	-
ABC.com	Jan-2018	0.25	-	(300 * (1 + .25)) = 375
ABC.com	Feb-2018	0.1	-	(375 * (1 + 1.1)) = 412.5
ABC.com	Mar-2018	-0.05	-	(412.5 * (1 - 0.05)) = 391.9
ABC.com	Apr-2018	0.3	-	(391.9 * (1 + 0.3)) = 509.4

Thank you!

[MK_QSL]

We can do this in script... but you need to provide sample data for that so that we can modify the script and reload the app.

[Anonymous]

Oh sorry about that, I have attached the excel file in the post. Thanks!

[MK_QSL · Accepted solution]

traffic:
LOAD domain, 
     date(num#(month_year),'MMM-YYYY') as month_year, 
     percentage
FROM
[Traffic Numbers.xlsx]
(ooxml, embedded labels, table is Sheet1);

join
LOAD domain, 
     date(num#(month_year),'MMM-YYYY') as month_year,
     [actual traffic]
FROM
[Traffic Numbers.xlsx]
(ooxml, embedded labels, table is Sheet2);

Final:
Load
 domain,
 month_year,
 percentage,
 [actual traffic],
 Num(If(domain <> Previous(domain),[actual traffic], Peek('projectedtraffic')*(1+percentage)),'#,##0.0') as [projectedtraffic]
Resident traffic
Order by domain, month_year;

Drop Table traffic;

[Anonymous]

Thank you very much Manish! It worked just fine.

[wdchristensen]

I have found that this type of problem lends itself nicely to a recursive solution. Unfortunately I have no clue how to code a recursive solution in QLIK so I typically do these operations in my load script. If you are pulling from a SQL Server database you could use TSQL code like the example below.

DECLARE @TblVar TABLE (tv_Domain nvarchar(15), tv_MonthYear date, tv_Percent float, tv_Traffic float)

INSERT INTO @TblVar

VALUES

('ABC.com','12/01/2017',1,300),

('ABC.com','01/01/2018',0.25,0),

('ABC.com','02/01/2018',0.1,0),

('ABC.com','03/01/2018',-0.05,0),

('ABC.com','04/01/2018',0.3,0)

;With TblVar_cte As

    (

Select tv_Domain,      tv_MonthYear, tv_Percent,   tv_Traffic

       FROM @TblVar

       WHERE tv_MonthYear = '12/01/2017'

Union All

Select cte.tv_Domain,  DateAdd(month,1,cte.[tv_MonthYear]) as tv_MonthYear,   tv.tv_Percent,       (tv.tv_Percent* cte.tv_Traffic) + cte.tv_Traffic as [tv_Traffic]

From TblVar_cte as cte INNER JOIN @TblVar as tv on DateAdd(month,1,cte.[tv_MonthYear]) = tv.[tv_MonthYear]

Where cte.tv_MonthYear < '5/01/2018'

)

SELECT *

FROM TblVar_cte

[sunny_talwar]

If you want to do this on the front end... you can try this

=Sum(TOTAL <domain> [actual traffic]) * exp(RangeSum(Above(Log(1+percentage), 0, RowNo())))

[wdchristensen]

Sunny, that is a slick formula you came up with! Will the solution you posted above work in Qlik Sense as well?

[sunny_talwar]

It should, but need to make sure that the sorting is correct in the chart.... or alternatively dictate sorting using Aggr() function

The sortable Aggr function is finally here!