time_weights: Weights for timing

In labelrank: Predicting Rankings of Labels

Description

This function calculates the diminishing weights for label ranking probabilities in case of timing nature of rankings.

Usage

time_weights(x, n)

Arguments

x	a scalar of timing periods.
n	is a parameter of 'memory' of how fast the past gets forgotten.

Details

Sometimes rankings have a 'timing' component (for example, weekly sport teams standing) and a recent event can be more important than the past. The model can take advantage of this difference in importance by weighting the ranking probabilities. The weights are calculated using an exponential function . In case of n=1, weights are a unitary vector of length n; thus, no time nature in rankings.

Value

a vector of values.

labelrank documentation built on May 2, 2019, 11:10 a.m.