vwp_multinomial_regression: Volleyball Win Probability Multinomial Regression Model

In dpwynne/volleyWP: Volleyball Win Probabilities

Description

Builds a model to predict the next attack from a touch and outputs the coefficient matrix

Usage

vwp_multinomial_regression(plays, formula, nmax = 2e+05, seed = 100)

Arguments

plays	a dv_plays object or data frame containing play-by-play data.
formula	a formula whose left-hand side contains the name of the variable in 'plays' containing the next attack, and whose right-hand side contains the features used in the regression
nmax	the maximum number of rows to use when fitting the model. Very large datasets can cause memory issues. If you are getting error messages related to an inability to allocate memory, set nmax to a smaller number. By default we use 200000.
seed	a seed for reproducibility of randomly sampling very large datasets.

Value

A coefficient matrix for a multinomial regression model, where each column represents one possible next attack and each row represents the contribution to the odds ratio of that outcome vs. the team winning the point without any more attacks made. The matrix has dimensions p x 8, where p is the number of features (including dummary variables) in the model matrix.

dpwynne/volleyWP documentation built on Dec. 20, 2021, 1:13 a.m.