The baseball.computer from droher

Nuances to handle:

Pickoffs should be credited to one and only one of the pitcher/catcher (as opposed to SB/CS), which requires correlating assist/putout data on the play
Rare/erroneous/missing details: pitcher unassisted, no fielding credit, or non-pitcher-catcher getting first assist (likely erroneous or P/C implied in pitch sequence). Can find with rg -e "PO\d\([3-8]"
Notify Retrosheet of box score generation error: https://www.baseball-reference.com/boxes/SEA/SEA201107170.shtml ("Pickoffs: Blake Beavan (0; Adrian Beltre, 2nd base)")

Pitching IP rate stats divided by (outs_recorded * 3) instead of (outs_recorded / 3)

I haven't been able to find the root cause of the problem, but I've stumbled across some columns of incorrect data. So far, the most notable example of this I can find is ERA in the metrics_player_career_pitching table. I've provided a quick example below with Terry Adams, but from what I can tell, many (if not all) pitchers have incorrect stats for multiple columns.

Querying some basic data in metrics_player_career_pitching for Terry, this is what gets returned:

Double-checking with baseballreference, Terry did not have a career ERA of 0.64, it was actually 4.17. His strikeouts/9 are also wrong, with an actual value of 7.2, as well as his RA(4.64) and his WHIP (1.46). Notably, all the counting stats seem to be accurate as far as I can tell, but many of the rate stats (or in general, numbers stored as doubles) are off.

Running list of tables/calcs

Carry enums over from Rust once duckdb has ALTER TYPE for custom types

  enum_lookups:
    side: game.game_team
    doubleheader_status: game.game
    time_of_day: game.game
    game_type: game.game
    sky: game.game
    field_condition: game.game
    precipitation: game.game
    wind_direction: game.game
    entered_game_as: game.game_lineup_appearance
    frame: event.event
    baserunner: event.event_starting_base_state
    baserunning_play_type: event.event_baserunning_play
    contact: event.event_batted_ball_info
    general_location: event.event_batted_ball_info
    depth: event.event_batted_ball_info
    angle: event.event_batted_ball_info
    strength: event.event_batted_ball_info
    fielding_play: event.event_fielding_play
    flag: event.event_flag
    sequence_item: event.event_pitch
  enum_vals:
    base: [First, Second, Third, Home]

{% macro add_enums() %}
  {% set sql%}
    {% for enum, table in var("enum_lookups").items() %}
      CREATE TYPE {{ enum }} AS ENUM (SELECT DISTINCT {{ enum }} FROM {{ table }});
    {%- endfor %}
    {% for node in graph.sources.values() %}
      {% for col_name, col_data in node.columns.items() if col_data.get("data_type") %}
        ALTER TABLE {{ node.schema }}.{{ node.name }} ALTER COLUMN {{ col_name }} TYPE {{ col_data.data_type }};
      {%- endfor %}
    {% endfor %}
  {% endset %}
  {% do log(sql, info=True)%}
  {% do run_query(sql) %}
{% endmacro %}

droher / baseball.computer Goto Github PK

baseball.computer's People

Contributors

Stargazers

Watchers

Forkers

baseball.computer's Issues

Fill out event skeleton for box score games

Add config to count lines of SQL in Github repo stats

Create separate table/subquery to properly handle pickoffs

Pitching IP rate stats divided by (outs_recorded * 3) instead of (outs_recorded / 3)

Running list of tables/calcs

Carry enums over from Rust once duckdb has ALTER TYPE for custom types

Recommend Projects

React

Vue.js

Typescript

TensorFlow

Django

Laravel

D3

Recommend Topics

javascript

web

server

Machine learning

Visualization

Game

Recommend Org

Facebook

Microsoft

Google

Alibaba

D3

Tencent