Correct Data Structure for Collection with Multiple Mutable Look-Up Properties

Question

If I have an in memory collection of 5,000 or more objects of type SampleObject

class SampleObject {
  ...
  Date getLastRefreshDate()
  Status getCurrentStatus()
}

I want to quickly get the sub-list of objects with a refresh date that is older than some value, and also be able to quickly get the objects with a certain status, what data-structure/algorithm would be useful? Is iterating the list and doing comparisons fast enough? What if the list grows to 25,000 or more?

Answer 1

A TreeMap<Date, SampleObject> could do the job of getting "older than" a certain date, pretty easily -- you'd just use headMap to get all the objects older than some value.

You'd need a separate Map<Status, List<SampleObject>> (or, if you can use third-party libraries, a Guava Multimap ) to track objects with some particular status, though, but I think a second data structure is inevitable anyway if you're not willing to pay for linear search.

Answer 2

The NavigableSet and NavigableMap classes offer methods to do just that.

NavigableSet already offer methods like headSet and tailSet to get all elements before or after a other given element. You could use your date criteria as a Comparator if not already povided as a natural order for your SampleObject class.

Besides other useful methods like lower , floor , ceiling , and higher .

Likewise NavigableMap offer similar methods like headMap and tailMap to do exactly the same kind of slicing.

Answer 3

get the sub-list of objects with a refresh date that is older than some value, and also be able to quickly get the objects with a certain status

Sounds like a k -dimensional range or search problem. Your options include:

• a kd -tree
• a range tree

You might be able to get away with lineary access via a sorted structure, if you mostly return data in one dimension only.