Is this an error in documentation of "numeric_detection"

Elastic Stack Elasticsearch
1

Hi,
I was going through the documentation of Dynamic Field Mapping.

In the table I see the last entry:

JSON data type.......................................Elasticsearch data type
string........................................................ Either a date field (if the value passes date detection), a double or long field (if the value passes numeric detection) or a text field, with a keyword sub-field.

My understanding is that the default will be double or long.

However in the same page later on, the section on Numeric Detection has an example where it is said:

The my_float field is added as a float field.

It mapped it as a float instead of double.

2

It depends on what is detected.

Again, it depends on what is detected.

3

Hi Mark,

Here is what the documentation is saying on the numeric types:
long -> A signed 64-bit integer
double -> A double-precision 64-bit IEEE 754 floating point number
float -> A single-precision 32-bit IEEE 754 floating point number

And it is here I am getting confused. It should have detected it as a long or double. But it detected it as float.

Should the documentation for Dynamic field mapping say that it can be detected as long, double or float?

4

It's picked the smallest "size" mapping for the provided my_float value, which is a float.

5

The documentation is saying:
string -> Either a date field (if the value passes date detection), a double or long field (if the value passes numeric detection) or a text field, with a keyword sub-field.

There is no mention of the fact that it will strive to choose the smallest "size".

I know it might look like splitting the hair but I am getting tripped with error
Bad Request: [status_exception] cannot merge mappings because of differences for field
while creating a Analytics Job.

The data is monthly data and I see that for some months the data is mapped a double and for others as float. Hence the question.

6

If you are using dynamic mapping then there may be some differences as Elasticsearch tries to do its best, but it's not guaranteed to be consistent.

Sounds like you should use a mapping template instead.

7

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.