Eliminate decimals past the thousandths place in a table

I have a table being pulled from a database that contains 'string' and 'double' data that being a parameter name and it's associated limits. I am currently trying to eliminate replicate data using the unique function however some of the data that is considered duplicated is not removed because it has either 14 or 16 trailing zeros. I would like to eliminate the zeros past the thousandths place.

Fix, Ceil, Round, and Floor do not work with tables.

[James Tursa]

Please upload a sample of the actual data you are working with, along with your exact desired output for this data.

[Adam Danz]

If those values are considered unique by the unique() function that eliminates the possibility that you're just using a long format such as

format long

but if those values are numeric, it suggests that they have some trailing non-zero decimal values, otherwise they wouldn't end in a string of 0s.

You could round all of your values to the thousandths place by using

round(x,3)

[Image Analyst]

Adam, can you post it down below in the Answers section?

Answer 1

[Adam Danz]

If those values are considered unique by the unique() function that eliminates the possibility that you're just using a long format such as

format long

but if those values are numeric, it suggests that they have some trailing non-zero decimal values, otherwise they wouldn't end in a string of 0s.

You could round all of your values to the thousandths place by using

round(x,3)

[Guillaume]

I think that the issue may be that camicat doesn't know how to apply round to the numeric variables of his table.

If so, the easiest may be:

numericvars = varfun(@isnumeric, yourtable);
yourtable(:, numericvars) = varfun(@(var) round(var, 3), yourtable, 'InputVariables', numericvars);

If all the variables of the table are numeric the above can be simplified to:

yourtable(:, :) = varfun(@(var) round(var, 3), yourtable);

[camicat]

Hi all, thank you for your help thus far!

This is an example of what I have on the table now. I have already passed it through the unique function sorting for any unique row of values and yet variable 1 and variable 3 have seemingly identical rows, with the only detectable difference being the number of trailing zeros between batches. For my purposes, I only really care about digits until the thousandths place.

Whenever I use the round function originally I received the following error:

"First argument must be a numeric logical or char array."

This was the part of the code used:

TableName = round(TableName(:,3:end),3);

When I tried the anonymous function above by Guillaume, I received the following error:

" A table variable subscript must be numeric array containing real positive integers..."

This is what I would like the final table to look like. I cannot sort by Variable because there are variables with different row values.

[Adam Danz]

My guess is that you're using C = unique(A,___,'rows') which treats each row as a single entity (see the docmentation). That won't return the unique values in each column. The reason Variable 1 has 2 matching values in column 3 is because it has unique values in column 1. Same with Variable 3.

Are you trying to list the unique values in each column? That will result in a different number of rows for each column which means you'd need to store the results in a cell array (1 element per column) or you would have to user NaNs are fillers so each column has the same number of rows.

Example

T = table({'X1';'X2';'X3'},[1.9;1.9;1.0],[2;2;2]); 
colUnique = varfun(@unique,T,'OutputFormat','cell');

[camicat]

Hi Adam!

I'm using unique(table(:,3:end),'rows') which would omit the first 2 columns when going through the table. It works for 98% of the data except the two batches I talked about where I come across an issue. Even if the values for those rows have non-zero decimal values past the trailing zeros as you suggested, for my purposes they are considered the same if they have the same number to the thousandths place, which is why I need to round the numbers or chop them after the thousandths place.

I am trying to list unique groups of values by row, not by column, so each row has the same number of columns.

Thank you for your help thus far!

[Walter Roberson]

uniquetol()

[Adam Danz]

Yeah, I'd try Walter's suggestion first by replacing the unique function my previous comment with uniquetol. If that doesn't work, you could play around with tolerance level. See C = uniquetol(A,tol)

If all else fails, you can round that column of data using

T.Var = round(T.Var,3); 

where T is your table and var is the column's variable name.