search

lukejharmon / traitathon

for storing stuff associated with the trait group at the tree-for-all hackathon
MIT License
2 stars 1 forks source link

arbor table handling #23

Open chodon opened 10 years ago

chodon commented 10 years ago

when you try to use aggregate table by average, if there is nothing entered in a cell, the function won't work for that column for the species, it still works in columns where there is not empty data. If you remove all empty cells in the column, it will run.

chodon commented 10 years ago

also, when you upload a csv file, sometimes the header rows are recognized, sometimes they are not...

curtislisle commented 10 years ago

agreed on CSV import problems. I hate it when the CSV upload gets hosed. I like posting problems like this as issues here. We can work out changes to the Arbror analysis. I am not sure why, exactly the upload problems occur. Is the offending dataset you are working with on the traitathon github repo? Or can you email it. I had run the aggregate by average on a previous version of the heliconia morph matrix, but maybe never noticed the empty cell error. I'll look at the aggregation problem first.

chodon commented 10 years ago

yes, we might have noticed this before because the previous versions of the heliconia morph matrix had artificial 0s instead of NAs for missing data. I can't imagine why I would have turned them into 0s except to mask the problem during some earlier iteration of the function? It could have been at the 1st hackathon that I did this! that was years ago! I don't know if it works on the csv's with NAs entered, I just know it doesn't work on csv's with empty values. I will email you some example files for both the aggregation and the column header issues (just wanted to continue the discussion here, to keep track of it in one place).

curtislisle commented 10 years ago

Sounds like a plan. Thanks for trying the aggregations. They were adjusted to handle NA values, which I guess R doesn't like.. I'll look at making the Arbor method tolerant of both NA and empty cells. I'll test with data I receive from you later. Thanks again.

lukejharmon commented 10 years ago

I am working on the aRbor code to better handle missing data. R is inconsistent with this sort of thing, but it’s possible to deal with it on a case-by-case basis.

On Sep 17, 2014, at 9:00 AM, Curtis Lisle notifications@github.com<mailto:notifications@github.com> wrote:

Sounds like a plan. Thanks for trying the aggregations. They were adjusted to handle NA values, which I guess R doesn't like.. I'll look at making the Arbor method tolerant of both NA and empty cells. I'll test with data I receive from you later. Thanks again.

— Reply to this email directly or view it on GitHubhttps://github.com/lukejharmon/traitathon/issues/23#issuecomment-55916302.

curtislisle commented 10 years ago

Aggregations have been fixed to work with empty cells and updated in public Arbor. Aggregation by max outputs internal max value (9E20) when there is no data present. Will fix this someday..