Skip to contents

Prepare PBP data for xG model predictions

Source: R/helpers_nhl.R
helper_nhl_prepare_xg_data.Rd

Helper function to prepare fastRhockey PBP data for expected goals (xG) calculations.

Adapted from the hockeyR helper_nhl_prepare_xg_data() function. The feature engineering logic (lag events, era dummies, shot metrics, one-hot encoding) is identical to the hockeyR xG training pipeline so that the same xgboost models can be used for inference.

Usage

helper_nhl_prepare_xg_data(x)

Arguments

x

A play-by-play data frame produced by nhl_game_pbp or nhl_game_feed.

Value

A tibble with model features plus identifiers (season, game_id, event_id, strength_state).

Details

The function:

  1. Filters out shootouts, penalty shots, and shift-change events.

  2. Computes lag features (last event type/team, time since last, distance from last, event zones).

  3. Adds era dummies for the four training eras (2011-13, 2014-18, 2019-21, 2022+).

  4. Computes tactical booleans: rebound, rush, cross_ice_event.

  5. One-hot encodes shot_type and last_event_type.

  6. Pads any missing model features to zero so the xgboost DMatrix dimensions match the trained models.