Hello.
Found information about a bug "Calculated field working on preview, but not in analysis " here and here, but the last message was in June 2023, is there any update on the topic? I have the same issue.
Hi,
I do not see this behaviour happening. When I do not have a calculated field using parseInt I get all the rows
And when I add a calculated field ifelse({col1}='a', parseInt({col2}), null) the lines that cannot be parsed are not shown.
From the community posts, I think the problem was that in the dataset preview it was showing the values and then not in the analysis?
Can you please elaborate your specific use case and issue? Thanks!
Sure, thank you for the answer.
I have calculated field like:
ifelse(
isNotNull({column_1}), {other_column_1},
isNotNull({column_2}), {other_column_2},
‘some_value’
)
On dataset level I have correct result - other_column_1 if column_1 is not Null;
other_column_2 if column_2 is not Null;
‘some_value’ if column_1 and column_1 is NULL.
But in the analysis I have completely different field.
After some research i found out, that if I use another calculated field like:
ifelse(
{column_1} <> “”, {other_column_1},
{column_1} <> “”, {other_column_2},
‘some_value’
)
at works totally fine.
So, I guess, the issue is resolved, but I’m confused about why it has happened.
P.S. column_1 and column_2 were created from json field using custom SQL query.
Hi,
I tried this and I see no difference. My calculated field is
ifelse(
isNotNull({col1}),{col1_1},
isNotNull({col2}),{col2_1},
'some_text'
)
What do you see when you move from the data prep to the analysis?
Unfortunately, I cannot share the direct screenshots.
In dataset Column1 and Column2 are empty and look like this:
Calculated field in dataset looks like this:
So, if Column1 and Column2 are Null, calculated field is “some_value” and it’s what i have expexted.
But in analysis I see this:
But if I’m using {column_1} <> “” instead of isNotNull({column_1}) everything works fine.
That is strange, becuase I am using the isNull and it shows some_text in both the data preparation and the analysis.
Is it possible to create a dataset, with some mock data in Arena that shows this behaviour so I can take a look at what can be happening?
I created this dashboard in Arena with a dataset that has empty fields, not (null), and the output is what I expect, both in data prep and on the dashboard, the isNotNull takes the (empty) value as not null and then shows an “A”, if I use the formula with “” as you mention, then the result is correct since the value of the cells is not (null) it is (empty)
What I am not sure is how is it showing (null) on your analysis if it is (empty) in your data set.
Hi @Valentin, I am marking this reply as, “Solution,” in case you need further assistance with your problem, please create a sample dashboard with sample dataset showing your problem state using Arena and please create a new post, so we can look at it from a fresh perspective. (Details on using Arena can be found here - QuickSight Arena)