arules (version 1.5-4)

hits: Computing Transaction Weights With HITS

Description

Compute the hub weights of a collection of transactions using the HITS (hubs and authorities) algorithm.

Usage

hits(data, iter = 16L, tol = NULL,
     type = c("normed", "relative", "absolute"), verbose = FALSE)

Arguments

data

an object of or coercible to class transactions.

iter

an integer value specifying the maximum number of iterations to use.

tol

convergence tolerance (default FLT_EPSILON).

type

a string value specifying the norming of the hub weights. For "normed" scale the weights to unit length (L2 norm), and for "relative" to unit sum.

verbose

a logical specifying if progress and runtime information should be displayed.

Value

A numeric vector with transaction weights for data.

Details

Model a collection of transactions as a bipartite graph of hubs (transactions) and authorities (items) with unit arcs and free node weights. That is, a transaction weight is the sum of the (normalized) weights of the items and vice versa. The weights are estimated by iterating the model to a steady-state using a builtin convergence tolerance of FLT_EPSILON for (the change in) the norm of the vector of authorities.

References

K. Sun and F. Bai (2008). Mining Weighted Association Rules without Preassigned Weights. IEEE Transactions on Knowledge and Data Engineering, 4 (30), 489--495.

See Also

Class transactions, function weclat

Examples

Run this code
# NOT RUN {
data(SunBai)

## calculate transaction weigths  
w <- hits(SunBai)
w

## add transaction weight to the dataset
transactionInfo(SunBai)[["weight"]] <- w
transactionInfo(SunBai)

## calulate regular item frequencies
itemFrequency(SunBai, weighted = FALSE)

## calulate weighted item frequencies
itemFrequency(SunBai, weighted = TRUE)
# }

Run the code above in your browser using DataCamp Workspace