topic Re: Classify more than 32 quantiles in Data Management Questions

Classify more than 32 quantiles

DitlevReventlow — Fri, 03 Jul 2015 13:14:13 GMT

Hey everyone.

Is there anyway to classify with more than 32 quantiles? 32 quantiles is apparently the maximum that I can pick in Symbology-->classified--> classify.

Thank you.

Best regards

Ditlev Reventlow.

Re: Classify more than 32 quantiles

DanPatterson_Retired — Fri, 03 Jul 2015 13:35:10 GMT

Not sure, but sometimes these things are set in the AdvancedArcmapSetting.exe found in your system path if normally installed C:\Program Files (x86)\ArcGIS\Desktop10.2\Utilities

Re: Classify more than 32 quantiles

DitlevReventlow — Fri, 03 Jul 2015 14:29:39 GMT

Thanks, I tried to look in there, but I couldn't find anywhere to change it. I don't think it is possible in there.

Re: Classify more than 32 quantiles

XanderBakker — Fri, 03 Jul 2015 15:26:40 GMT

Can you elaborate a little on what you want to achieve with the classification of more than 32 classes? If it is for visualization, it might be difficult to distinguish 32 or more classes. Perhaps you can create a new field (or raster, depending on the data you are using) of type integer, that can be visualized with unique values (the values that represent the more than 32 classes).

Re: Classify more than 32 quantiles

DitlevReventlow — Fri, 03 Jul 2015 15:35:11 GMT

It is because I have a precipitation raster layer where values are just about normally distributed. I want to leave out e.g. 2,5 percent in each "tail" of the data distribution. This is because this data is very spread out and doesn't have any use value for me.

Using the 32 quantililes I can not mark and leave out less than 100/32=3,125 % in each tail of the distribution, so 6,25 % for both tails together. I would like to include a little bit more than 93,75% of the data - at least 95 %.

Re: Classify more than 32 quantiles

XanderBakker — Fri, 03 Jul 2015 15:39:48 GMT

If you classify manually, you can set the values of the tails exactly... do you need to have 30 classes between the "tails"?

Re: Classify more than 32 quantiles

DitlevReventlow — Fri, 03 Jul 2015 15:45:18 GMT

No, I don't need the 30 in between. These are just to insure that the intervals at the tails are sufficiently small (3,125%).

But I I classify manually then I guess there is no way for me to know how many percentage I leave out in each tail? And how do I make sure that I leave out the same percentage in each tail?

Re: Classify more than 32 quantiles

DanPatterson_Retired — Fri, 03 Jul 2015 16:13:54 GMT

If you know the mean and standard deviation why don't you use the mean and standard deviation classification that is built in and alter the tail colours there?

Re: Classify more than 32 quantiles

DitlevReventlow — Fri, 03 Jul 2015 17:00:13 GMT

Dan, I am sure I understand what you mean, could you decribe it further?

Re: Classify more than 32 quantiles

DitlevReventlow — Fri, 03 Jul 2015 17:00:37 GMT

*not sure I understand

Re: Classify more than 32 quantiles

XanderBakker — Fri, 03 Jul 2015 17:36:17 GMT

The alternative would be to create a new raster using Slice—Help | ArcGIS for Desktop with EQUAL AREA. This excepts more than 32 classes. If you use 40, you will have 2.5 percentile for each value produced. With 50 zones, each output value represents 2%, etc:

Re: Classify more than 32 quantiles

DitlevReventlow — Fri, 03 Jul 2015 19:05:35 GMT

I tried to use Slice and did this into 40 equal area intervals. This worked but the original value (the precipitation value) was lost. However, then I did combined it with the original raster which yielded a raster dataset where all the precipitation values have a number. Then I can use Con("rasterdataset"!=1,"rasterdataset") and Con("rasterdataset"!=40,"rasterdataset") to remove the two 2,5 % tails. Should solve the problem!

Doed this sound reasonable? I dont know if it could be solved it a smarter way, though.

Re: Classify more than 32 quantiles

XanderBakker — Sun, 05 Jul 2015 01:39:39 GMT

You could use the SetNull tool to set the tails to NoData. However, if the idea is just for visualization, you can better use suggestion Dan provided.