Microstation To Find Line Or Arc Of Best Fit.

Unknown said:

After browsing the functionality I have struggled to come up with anything. I am currently using Microstation Select Series 2.

Have you checked all Curve Functions Toolbox from  Microstation?

There is a Curve by Equation tool, also, but no help in the help file

I used  curve by equation tool a few times back in J and it only used to plot a curve given the equation not find the equation but it may have changed I'll have to go back and take a look .. might even have an old J training book lying around

Very cool!

I tried running it with just 9 points but it doesnt seem to work for me

do you put all the active points into a selection set then run the macro?

it seems to always put the answer in the same spot... no matter where I have my points...

hmmmmm

I just realised its linear only but it still does it in the same spot  with new points as crosses..

If it was working from selection then Id expect it to land near original points ...hmmm looking at the code I'm no closer to why?

Unknown said:

It seems to always put the answer in the same spot

The VBA example provides its own data, with the GetData method.  It's not looking for any other data.

Unknown said:

Do you put all the active points into a selection set then run the macro?

I made no claim about active points or selection sets.  The example illustrates a programmatic solution to the original question.  If you want to collect data from another source, it's up to you to modify the code to achieve that.

[Edit]

Stefan Bernsdorf has kindly added code to use a selection set of MicroStation point objects.  His additions are included in the modified VBA project available here.

Unknown said:

It seems to always put the answer in the same spot

The VBA example provides its own data, with the GetData method.  It's not looking for any other data.

Unknown said:

Do you put all the active points into a selection set then run the macro?

I made no claim about active points or selection sets.  The example illustrates a programmatic solution to the original question.  If you want to collect data from another source, it's up to you to modify the code to achieve that.

[Edit]

Stefan Bernsdorf has kindly added code to use a selection set of MicroStation point objects.  His additions are included in the modified VBA project available here.

ok that would explain it ... thanks still pretty cool bit of coding .. need to study more...

Hi Lorys,

anyone mind if I add the desired functionality ;-)

Regards, Stefan.

no go ahead please... but its still only linear regression.....

it would be really nice if it could determine linear or logarithmic from correlation coefficient then use the best fit equation for the type of curve like excel does  or like I  did in statistical analysis back in college before excel could do it we did it with Texas instrument basic ... I could find the equations ( still have my old text book)  but I no longer have the program code I wrote for basic  which asked for keyin points then run if linear gave the R squared  value  and if logarithmic gave the R squared  value .. and  you then decided the closest to one  is the type of curve and it then gave you the equation of the curve.... and I wrote all that on my Commodore 64 ( C64) with floppy disk drive for my final project at college to determine the data from experimental results ... it sure was fun learning the math then applying it to TI Basic  to solve real problem and it saved me hundreds of hours calculations for 25 different experimental trials ..

Excel makes it too easy now...arent we all glad of that!

Unknown said:

It would be really nice if it could determine linear or logarithmic from correlation coefficient then use the best fit equation for the type of curve like Excel does

Perhaps it would be nice, but consider that this question is the first time (in two decades) that anyone has posted about linear regression.

Numerical analysis and regression algorithms have been around for over a century.  Tools that are commonly used to manipulate arbitrary sets of data implement those algorithms.  Excel is a good example, which is why Microsoft invested a lot of development time to create those tools you endorse. 

Conversely, MicroStation is not commonly used to analyse random sets of data.  One question in two decades about linear regression is not a big incentive to encourage anyone to put significant development effort into writing a similar set of tools for MicroStation.  Find me one hundred customers and prove me wrong!

Unknown said:

One question in two decades about linear regression is not a big incentive to encourage anyone to put significant development effort into writing a similar set of tools for MicroStation

That's why not so lot "Land Surveyors" are using plain Microstation :-),

but it's a common method for Surveying industries working with survey data.