Ordinal Value Prediction Task
(Redirected from ordinal prediction task)
Jump to navigation
Jump to search
An Ordinal Value Prediction Task is a rank decisioning task that is a prediction task (which requires an ordinal value).
- Context:
- Input: an item (often a data item); and a rank set.
- output: one or more rank set members (possibly as rank member identifiers).
- measures: Ranking Task Performance Measures.
- It can range from being a Binary Ranking Task to being a Multirank Ranking Task (such as large multirank ranking), depending on the rank set size.
- It can range from being a Unirank Ranking Task to being a Multirank Ranking Task, depending on the number of rank outputs,.
- It can range from being a Simple-Input Ranking Task to being a Complex-Input Ranking Task, depending on the complexity of the input item.
- It can range from being a Model-based Ranking Task to being ...
- It can range from being a Manual Ranking Task to being an Automated Ranking Task.
- It can range from being a Heuristic Ranking Task to being a Data-Driven Ranking Task (such as a supervised ranking), depending on whether some uncertainty is allowed in the decision.
- It can range from being a Probabilistic Ranking Task to being ...
- It can range from being a Batch Ranking Task to being an Online Ranking Task.
- It can be solved by a Ranking System (that implements a ranking algorithm).
- It can be solved by an Ordinal Value Prediction System (that implements an ordinal value prediction algorithm).
- It can be instantiated as a Ranking Act.
- It can have a Cost Function with Ranked costs.
- Example(s):
- Predict whether a competitor will come in 1st place, 2nd place, 3rd place, or other.
- Predict whether a person will rank a movie as excellent, good, okay, boring, bad.
- INEX.
- …
- Counter-Example(s):
- See: Ranking Task, Ordinal Function, Item Recommendations System Performance Measure.
References
2006
- (Li & Link 2006) ⇒ Ling Li, and Hsuan-Tien Lin. (2006). “Ordinal Regression by Extended Binary Classification.” In: Advances in Neural Information Processing Systems 19 (NIPS 2006).
- QUOTE: We work on a type of supervised learning problems called ranking or ordinal regression, where examples are labeled by an ordinal scale called the rank. For instance, the rating that a customer gives on a movie might be one of do-not-bother, only-if-you-must, good, very-good, and run-to-see. The ratings have a natural order, which distinguishes ordinal regression from general multiclass classification. … In an ordinal regression problem, an example [math]\displaystyle{ (\mathbf{x}, y) }[/math] is composed of an input vector [math]\displaystyle{ {\bf x} \in \mathcal{X} }[/math] and an ordinal label (i.e., rank) [math]\displaystyle{ y \in \mathcal{Y} = {1, 2,...,K} }[/math]. Each example is assumed to be drawn i.i.d. from some unknown distribution [math]\displaystyle{ P({\bf x}, y) }[/math] on [math]\displaystyle{ \mathcal{X} \times \mathcal{Y} }[/math].