Comments on: LOESS Utility – Awesome Update

By: Jon Peltier

Jon Peltier — Mon, 01 Feb 2010 02:41:18 +0000

Nick – I’m preparing some documentation for a new article. Stay tuned.

By: Nick

Nick — Mon, 01 Feb 2010 00:22:01 +0000

Hi Jon!
Nice little utility you have here! I’m utterly hopeless at math and I have a report involving normalizing microarray data, which requires me to apply LOESS to some data. I have trouble following your example of plotting LOESS for anomaly data for NASA. There are seven choices for entering the data and I don’t know which to choose! If you give a step by step instruction of how you used the add in, it would be great.
Please help the mathematically crippled biologist!

By: Jon Peltier

Jon Peltier — Sat, 30 Jan 2010 14:36:05 +0000

Tom -

I don’t know what may have caused your first error. When it was reported last time, the problem was resolved by selecting just the data, not the entire column. I have added a slightly different dange auto-detect feature. Select the top cell in a range of data (the first data cell, not the header), and the utility will detect how far down the column the range extends.

I also don’t know about the second crash. Similar problems arise if not all upgrades and service packs have been installed.

How frequently does the dialog vanish? I see it once in a while, perhaps once every few dozen times I run the utility. I’ve just assumed it was related to the inherent instability of the RefEdit used to select the data ranges.

By: tom

tom — Fri, 29 Jan 2010 22:24:18 +0000

I also experienced the runtime error.
Does it have to do with very small alpha values? I chose 20 points out of 2000.

bug #2:
The utility crashes from the menu if Excel is opened without any workbook.

bug #3:
Sometimes the dialog disappears when I am entering ranges in the text boxes.

idea:
Could you add something like a auto-fit for columns in the tool. I rather would like
to add full columns. The utility could start the range below the last text row and
would end the range before the first empty cell.

By: Jon Peltier

Jon Peltier — Fri, 18 Dec 2009 01:23:46 +0000

Hi Omair -

I just did some fooling around, and had no problem with a data set with 5000 points. But I used some vanilla data, and yours may have some characteristic that makes it behave differently.

Is it possible to send me a sample? jonpeltier at-sign gmail period com.

By: Omair Noor

Omair Noor — Fri, 18 Dec 2009 00:14:44 +0000

Hi Jon,

I have approximately 2400 data points per column. I tried to use your LOESS utility function, but i get the following error:

Run-time error ‘6′:

Overflow.

Any idea how should i be going about resolving this problem.

Thanking in anticipation.

By: Jon Peltier

Jon Peltier — Mon, 12 Oct 2009 14:43:55 +0000

The bare minimum is Excel 2000, but I’m sure that would be nearly impossible to find. The best version so far is Excel 2003, and even that isn’t easy to get your hands on.

By: Stephen Hughes

Stephen Hughes — Mon, 12 Oct 2009 14:33:27 +0000

Hi Jon,

Anything you can manage would be appreciated! Meanwhile I’ll try (again) to persuade my boss of the benefits of having a modern computer and up-to-date software!

By: Jon Peltier

Jon Peltier — Mon, 12 Oct 2009 14:26:18 +0000

Stephen -

There was a major upgrade of VBA between Excel 97 and Excel 2000, and Macs have never gotten this upgrade. I would expect all of my code to work fine in Excel 2000 and later, but I would expect problems in Excel 97 and Mac Excel.

I don’t know if I even still have Excel 97 disks. If I do, maybe I’ll install it in a VM and see where the code falls down.

By: Stephen Hughes

Stephen Hughes — Mon, 12 Oct 2009 12:40:35 +0000

Is there a minimum version of Excel needed to run this? I’ve got 97 but get a ‘Compile Error in Hidden Module: M_LOESS’ error when I try to run it.

Any suggestions would be appreciated as I need a program to smooth out experimental data.