BOPR: BOPR

In haven-jeon/BOPR: Bayesian online learning scheme for probit regression with R

Description

Bayesian online learning scheme for probit regression (BOPR)

Usage

BOPR(x, ...)

## Default S3 method:
BOPR(x, y, beta = 0.05, prior_prob = 0.5,
  epsilon = 0.05, subset = NULL, ...)

## S3 method for class 'formula'
BOPR(formula, data, subset = NULL, na.action = na.pass,
  beta = 0.05, prior_prob = 0.5, epsilon = 0.05, ...)

Arguments

x	a matrix of predictors.
...	not used
y	a factor vector with 2 level
beta	scaling parameter
prior_prob	prior of initial parametes
epsilon	parameter to apply dynamics
subset	optional expression saying that only a subset of the rows of the data should be used in the fit.(currently it's not working.)
formula	an optional data frame in which to interpret the variables named in the formula.
data	an optional data frame in which to interpret the variables named in the formula.
na.action	a function which indicates what should happen when the data contain NAs.

Value

S3 BOPR object; a list of consisting of

beta_matrix	beta matrix with mean and variance
beta	scaling parameter
prior_prob	prior of initial parametes
epsilon	parameter to apply dynamics
formula	formula

Author(s)

Heewon Jeon [e-mail]

References

Graepel, Thore, et al. "Web-scale bayesian click-through rate prediction for sponsored search advertising in microsoft's bing search engine." Proceedings of the 27th International Conference on Machine Learning (ICML-10). 2010. He, X., Bowers, S., Candela, J. Q., Pan, J., Jin, O., Xu, T.,Herbrich, R. (2014). Practical Lessons from Predicting Clicks on Ads at Facebook. Proceedings of 20th ACM SIGKDD Conference on Knowledge Discovery and Data Mining - ADKDD '14, 1-9.

Examples

idx  <- sample(1:nrow(credit_approval))
first_train_set  <- credit_approval[idx[1:200],]
second_train_set  <- credit_approval[idx[201:400],]
test_set <- credit_approval[idx[401:690],]

bopr_mdl <- BOPR(A16 ~ A1 + A4 + A5 + A7 + A9 + A10 + A12 + A13 , first_train_set)
bopr_mdl_up  <- online_leraning(bopr_mdl, second_train_set)
pred  <- predict(bopr_mdl_up, test_set)

haven-jeon/BOPR documentation built on May 17, 2019, 3:05 p.m.