BTdataframe: Dataframe initialization

In BTdecayLasso: Bradley-Terry Model with Exponential Time Decayed Log-Likelihood and Adaptive Lasso

Description

Dataframe initialization

Usage

BTdataframe(dataframe, home = TRUE)

Arguments

dataframe

Raw dataframe input, an example data "NFL2010" is attached in package for reference The raw data is a dataframe with 5 columns. First column is home teams. Second column is away teams. Third column is the number of wins of home teams (if home team defeats away team, record 1 here, 0 otherwise). Fourth column is the number of wins of away teams (if home team defeats away team, record 0 here, 1 otherwise). Fifth column is a scalar of time when the match is played until now (Time lag). Any time scale can be used here. "NFL2010" applies the unit of day.

home

Whether home effect will be considered, the default is TRUE.

Details

Initial the raw dataframe and return an un-estimated ability vector and the worst team who loses most.

Note that even if the tournament does not have any home team or away team, you can still provide the match results according to the description above regardless of who is at home and who is away. By selecting the home = FALSE, We duplicate the dataset, switch the home, away teams and also the home, away match results. Then this dataset will be attached to the original dataset and all home and away win's number will be divided by 2. MLE estimation of home effect is proved to be an exact 0.

The elimination of home effect by duplicating the original dataset will be less efficient than eliminating the home parameter directly in iterations. Since most games such as football, basketball have home effect and this method provides an idea of handling the case where some games have home effect and some games are played on neutral place, this method is applied here.

Value

dataframe	dataframe for Bradley-Terry run
ability	Initial ability vector for iterations
worstTeam	The worst team whose ability can be set as 0 during any model's run

BTdecayLasso documentation built on May 1, 2019, 8:24 p.m.