B_T_Weighted_Lasso_R_wij: Bradley-Terry Weighted Likelihood function with and Lasso...

In heilokchow/MWLE-Lasso: Bradley-Terry Model with Exponential Decayed Likelihood function and Lasso Penalty

Description

Bradley-Terry Weighted Likelihood function with and Lasso Peanlty's optimization Prediction on one future period is done

Usage

B_T_Weighted_Lasso_R_wij(df, ability, i0, date, u, penalty, dd, uk)

Arguments

ability	A column vector of teams ability, the last row is the home parameter whose rowname must be "at.home".
i0	A teams index whose ability will be fixed as 0 (usually the team loss most).
date	A vector which separate all rows of datasets into many parts.
u	The exponential decay rate.
penalty	Penalty level = s/max(s)
dd	The index of one value from date. Trained model will be run before this date and prediction will be done on the previous part of dataset.
uk	Exponetial rate between lambda and true penalty level (usually setting to be 0.642)
dataframe	Dataframe with 5 columns. First column is the index of the home teams (use numbers to denote teams). Second column is the index of the away teams. Third column is the number of wins of home teams (usually to be 0/1). Fourth column is the number of wins of away teams (usually to be 0/1). Fifth column is the scalar of time when the match is played until now (Time lag).
lambda	Lasso penalty of no statistical mean, a larger choice of lambda means higher penalty. Usually the best penalty is chosen from Cross-Validation, where in-model's prediction power is maximized.

Value

The estimated abilities and predicted likelihood

heilokchow/MWLE-Lasso documentation built on May 23, 2019, 4:03 a.m.