Storing and Using Value Ranges (PostgreSQL) - postgresql

Question

I have a PostgreSQL database where physical activities store a certain energy decimal value, e.g.
ACTIVITY ENERGY
-----------------
Activity1 0.7
Activity2 1.3
Activity3 4.5
I have a Classification system that classifies each Energy value as
Light: 0 - 2.9
Moderate: 3.0 - 5.9
Vigorous: >= 6.0
The Classification and Energy Values are subject to change. I need a way to quickly get the Type of each activity. But how to store these in a way which is easy to retrieve?
One solution, is define MIN/MAX lookups of the Type "Classification" -- and pull up all available classifications; then do a CASE/WHEN to go through each one.
LOOKUP_ID LOOKUP_NAME LOOKUP_VALUE LOOKUP_TYPE
---------------------------------------------------------
1 LIGHT_MIN 0 CLASSIFICATION
2 LIGHT_MAX 2.9 CLASSIFICATION
3 MODERATE_MIN 3 CLASSIFICATION
4 MODERATE_MAX 5.9 CLASSIFICATION
5 VIGOROUS_MIN 6 CLASSIFICATION
6 VIGOROUS_MAX null CLASSIFICATION
But this doesn't look very easy to me -- if a developer needs to get the current Classiication they'll have to step through different cases and compare them.
Is there a better strategy to capture these ranges, or is this the right one?

Answer 1

Use a range type
create table classification
(
description text,
energy numrange
);
insert into classification
(description, energy)
values
('Light', numrange(0,3.0,'[)')),
('Moderate', numrange(3.0, 6.0, '[)')),
('Vigorous', numrange(6.0, null, '[)'));
Then you can join those two tables using the <# operator:
select *
from activity a
join classification c on a.energy <# c.energy
The nice thing about the range type is that you can prevent inserting overlapping ranges by using an exclusion constraint
alter table classification
add constraint check_range_overlap
exclude using gist (energy with &&);
Given the above sample data, the following insert would be rejected:
insert into classification
(description, energy)
values
('Strenuous', numrange(8.0, 11.0, '[)'));

Answer 2

I don't think this is a great solution, but it seems preferable to the model above.
Create a table with the ranges and classifications:
create table classification (
energy_min numeric,
energy_max numeric,
classification text
);
Then do a join on that table as follows:
a.activity, a.energy, c.classification
from
activities a
left join classification c on
a.energy >= c.energy_min and
(a.energy <= c.energy_max or c.energy_max is null);
If the possible classifications is relatively small, this should work well enough. I don't think it's efficient on the back end, as it's likely doing a cross-join on the classification table. That said, if it's three (or even ten) records, it's not that big of a deal.
It should scale very well and enable you to modify values on the fly and get results quickly.
If you really want to get fancy, you can also include effectivity from and thru dates on the "classification" table that enable you to change the classifications but also retain historical classifications for older records.