Linest_m problem

[#1]

Hi All,

I want to calculate when (date) the lower line reaches the upper line in the graph attached by drawing a reference line or something. I understand i will have to use the linest_m function but cant figure out how to.

Thanks for your time.

[swuehl]

Hi ggoyal,

you might use the split axis (on axis tab) for that, but I wasn't able to get it working for your example.

Then, you probably need a trendline (general tab) and you need to forecast your trendline (axis tab).

You need to decide which trendline will be a good one for your data.

If this works out, you will just get a graph showing a trendline interecting  your reference line.

If you need a value (like trend hits reference July 2013), you might need to calculate that yourself.

Please have a look at the statistical aggregation fucntions in QV, e.g. linest_m and linest_n

Regards,

Stefan

[#3]

I think the first problem to tackle is how i can make the reference line and still make sure the rest of the data is being displayed correctly.

[swuehl]

What do you mean with data display correctly? I haven't noticed that your data is not being displayed correctly so far.

It easy to have both the reference line and your data on the same chart.

It sounds to me that you still want to see every little variation of your data in the scale of some 10 to thousend  GB while the reference line is 3 magnitudes larger than your data.

You could change the ordinate to logarithmic, that may help a little, though most people don't understand changes in logarithmic scale easily.

Or you could leave it as is, I think it is correct to show that your data is a bottom flat line relative to the reference. And as you said, before July 2013, this picture will change  and your data will have the same magnitute as your reference.

You could of course use a second axis for the reference, but I don't think this will help in understanding the data.

For example, your data might visually cross the reference, but only because scales are different.

One idea: you could add the reference as second expression, make displaying the second expression conditional by a button press and set axis to autoscale. By pressing the button, your users might switch between full picture and detailed picture.

Regards,

Stefan

P.S:

a) I have checked the spli axis thing. Works different than I expected. Don't think that this will help

b) do you really have a petabyte DB in use? Cool stuff...

[#5]

Can anyone please explain how to use the linest_m function?

[swuehl · Accepted solution]

Your data will cross the 32000 max DB size on 10/18/2011 using linear regression based forecast.

See the attached app and the text box expression for use of linest_m and linest_b. I think I don't need to go into linear regression, do I?

Instead of the fixed max DB limit you could of course enter a variable into reference line and text box expression.

Regards,

Stefan

[swuehl]

Hi ggoyal_123,

have you changed  the title / text of your postings?

Doing so will be break the history of our conversation, i.e. my first two postings now seems to be totally off topic regarding your question in your start of the thread and this makes it hard to understand what we have discussed earlier.

I recommend not to change any essential part of any conversation (with exception of minor corrections), at least make it clear if you changed something.

If the topic of your problem shifts just add new information to the thread or start a new thread.

Regards,

Stefan

[#8]

Thanks so so much for your time and effort! Yeah i had changed the post slightly becasue i wanted to attach a new QV file and didnt know how else to!