
Re: Positive/negative result from trendline
Jonathan Brough Dec 17, 2012 6:03 AM (in response to Filip Larsson)My problem is that these equations are builtin QlikView calculations, which I activate through the diagram proporties. Does anybody know how to access the calculation of the trend's direction (y value) so that I can show it in a text object.
Sorry, but I don't know what this means. Do you mean that the expression is set up somewhere other than in the chart properties ?
You could perhaps add another expression in a combo chart, test for increases or decreases in order to say something like 'Up' or 'Down', make sure each means of display (Bar, Line and Symbol) is deselected and then tick the 'Text on Axis' checkbox.
Jonathan

Re: Positive/negative result from trendline
Stefan Wühl Dec 17, 2012 6:15 AM (in response to Filip Larsson)Check out the linest_m function to calculate the slope of your linear regression:
From the HELP:
linest_m ([{set_expression}][ distinct ] [total [<fld {,fld}>] ]yexpression, xexpression[, y0 [, x0 ]])
returns the aggregated m value (slope) of a linear regression defined by the equation y=mx+b for a series of coordinates represented by paired numbers in xexpression and yexpression iterated over the chart dimension(s). Text values, null values and missing values in any or both pieces of a datapair will result in the entire datapair to be disregarded.
An optional valye y0 may be stated forcing the regression line to pass through the yaxis at a given point. By stating both y0 and x0 it is possible to force the regression line to pass through a single fixed coordinate.
Unless both y0 and x0 are stated, the function requires at least two valid datapairs to calculate. If y0 and x0 are stated, a single data pair will do.
This function has the same limitations for nested aggregation as the avg([{set_expression}] [ distinct ] [ total [<fld { , fld } >]] expression) function. The linest_m function supports Set Analysis and the total qualifier in the same way as the avg([{set_expression}] [ distinct ] [ total [<fld { , fld } >]] expression) function.
Examples:
linest_m( Y, X )
linest_m( A/B, X*Y/3 )
linest_m( total Y, X )
linest_m( total <Z> Y, X)
linest_m( Y, X, 0 )
linest_m( Y, X, 1, 1 )

Re: Positive/negative result from trendline
Filip Larsson Dec 20, 2012 5:06 AM (in response to Stefan Wühl )Thanks alot!
The linest_m function was exactly what I needed.
