FTRLProx_predict: FTRL-Proximal Linear Model Predicting

In yanyachen/rFTRLProximal: FTRL-Proximal Algorithm

Description

An advanced interface for FTRL-Proximal online learning model predicting.

Usage

FTRLProx_predict(model, data)

Arguments

model	a FTRL-Proximal linear model object.
data	a object of class ftrl.Dataset.

Value

an vector of linear model predicted values

Examples

library(data.table)
library(FeatureHashing)
library(MLmetrics)
data(ipinyou)
m.train <- FTRLProx_Hashing(~ 0 + ., ipinyou.train[, -"IsClick", with = FALSE],
                            hash.size = 2^13, signed.hash = FALSE, verbose = TRUE,
                            label = as.numeric(ipinyou.train$IsClick))
m.test <- FTRLProx_Hashing(~ 0 + ., ipinyou.test[,-"IsClick", with = FALSE],
                           hash.size = 2^13, signed.hash = FALSE, verbose = TRUE,
                           label = as.numeric(ipinyou.test$IsClick))
ftrl_model <- FTRLProx_train(data = m.train, model = NULL,
                             family = "binomial",
                             params = list(alpha = 0.01, beta = 0.1,
                                           l1 = 0.1, l2 = 0.1, dropout = 0), epoch = 50,
                             watchlist = list(test = m.test),
                             eval = AUC, patience = 5, maximize = TRUE,
                             nthread = 1, verbose = TRUE)
pred_ftrl <- FTRLProx_predict(ftrl_model, m.test)
AUC(pred_ftrl, as.numeric(ipinyou.test$IsClick))

yanyachen/rFTRLProximal documentation built on May 4, 2019, 2:30 p.m.