Skip to content

Commit

Permalink
refactored databallpy_events (#236)
Browse files Browse the repository at this point in the history 
* refactored databallpy_events

* updated version debugpy

* Updated docs
  • Loading branch information
@Alek050
Alek050 authored Aug 6, 2024
1 parent 0a16c3b commit 6103cdf
Show file tree
Hide file tree
Showing 27 changed files with 2,140 additions and 1,872 deletions.
43 changes: 24 additions & 19 deletions databallpy/data_parsers/event_data_parsers/metrica_event_data_parser.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -347,19 +347,17 @@ def _get_shot_event(
start_x=row.start_x,
start_y=row.start_y,
team_id=row.team_id,
pitch_size=pitch_dimensions,
team_side="home" if row.team_id == home_team_id else "away",
_xt=-1.0,
pitch_size=pitch_dimensions,
player_id=row.player_id,
jersey=players.loc[players["id"] == row.player_id, "shirt_num"].iloc[0],
shot_outcome=["miss", "goal"][row.outcome],
y_target=np.nan,
z_target=np.nan,
body_part=None,
type_of_play=None,
first_touch=None,
created_oppertunity=None,
outcome=bool(row.outcome),
related_event_id=MISSING_INT,
body_part="unspecified",
possession_type="unspecified",
set_piece="unspecified",
_xt=np.nan,
outcome_str=["miss", "goal"][row.outcome],
)


Expand Down Expand Up @@ -390,18 +388,22 @@ def _get_pass_event(
start_x=row.start_x,
start_y=row.start_y,
team_id=row.team_id,
pitch_size=pitch_dimensions,
team_side="home" if row.team_id == home_team_id else "away",
_xt=-1.0,
pitch_size=pitch_dimensions,
player_id=row.player_id,
jersey=players.loc[players["id"] == row.player_id, "shirt_num"].iloc[0],
outcome=["unsuccessful", "successful"][row.outcome]
outcome=bool(row.outcome),
related_event_id=MISSING_INT,
body_part="unspecified",
possession_type="unspecified",
set_piece="unspecified",
_xt=np.nan,
outcome_str=["unsuccessful", "successful"][row.outcome]
if not pd.isnull(row.outcome)
else "not_specified",
end_x=row.end_x,
end_y=row.end_y,
pass_type="not_specified",
set_piece="unspecified_set_piece",
pass_type="unspecified",
)


Expand Down Expand Up @@ -432,13 +434,16 @@ def _get_dribble_event(
start_x=row.start_x,
start_y=row.start_y,
team_id=row.team_id,
pitch_size=pitch_dimensions,
team_side="home" if row.team_id == home_team_id else "away",
_xt=-1.0,
pitch_size=pitch_dimensions,
player_id=row.player_id,
jersey=players.loc[players["id"] == row.player_id, "shirt_num"].iloc[0],
related_event_id=MISSING_INT,
duel_type=None,
outcome=bool(row.outcome),
has_opponent=False,
related_event_id=MISSING_INT,
body_part="unspecified",
possession_type="unspecified",
set_piece="unspecified",
_xt=np.nan,
duel_type="unspecified",
with_opponent=False,
)
105 changes: 57 additions & 48 deletions databallpy/data_parsers/event_data_parsers/opta_parser.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -8,7 +8,12 @@
from pandas._libs.tslibs.timestamps import Timestamp

from databallpy.data_parsers import Metadata
from databallpy.events import BaseOnBallEvent, DribbleEvent, PassEvent, ShotEvent
from databallpy.events import (
DribbleEvent,
IndividualCloseToBallEvent,
PassEvent,
ShotEvent,
)
from databallpy.utils.constants import MISSING_INT
from databallpy.utils.logging import create_logger
from databallpy.utils.tz_modification import convert_datetime, utc_to_local_datetime
Expand Down Expand Up @@ -109,9 +114,9 @@

SHOT_ORIGINS_QUALIFIERS = {
9: "penalty",
22: "regular_play",
22: "open_play",
23: "counter_attack",
24: "crossed_free_kick",
24: "free_kick",
25: "corner_kick",
26: "free_kick",
}
Expand Down Expand Up @@ -176,7 +181,9 @@

def load_opta_event_data(
f7_loc: str, f24_loc: str, pitch_dimensions: list = [106.0, 68.0]
) -> tuple[pd.DataFrame, Metadata, dict[str, dict[str, dict[str, BaseOnBallEvent]]]]:
) -> tuple[
pd.DataFrame, Metadata, dict[str, dict[str, dict[str, IndividualCloseToBallEvent]]]
]:
"""This function retrieves the metadata and event data of a specific match. The x
and y coordinates provided have been scaled to the dimensions of the pitch, with
(0, 0) being the center. Additionally, the coordinates have been standardized so
Expand Down Expand Up @@ -462,7 +469,7 @@ def _load_event_data(
away_team_id: int,
players: pd.DataFrame,
pitch_dimensions: list = [106.0, 68.0],
) -> tuple[pd.DataFrame, dict[str, dict[str | int, BaseOnBallEvent]]]:
) -> tuple[pd.DataFrame, dict[str, dict[str | int, IndividualCloseToBallEvent]]]:
"""Function to load the f27 .xml, the events of the match.
Note: this function does ignore most qualifiers for now.
Expand All @@ -478,7 +485,7 @@ def _load_event_data(
Returns:
pd.DataFrame: all events of the match in a pd dataframe
dict: dict with "shot_events", "dribble_events", and "pass_events"
as key and a dict with the BaseOnBallEvent instances
as key and a dict with the BaseIndividualCloseToBallEvent instances
"""

dribble_events = {}
Expand Down Expand Up @@ -639,7 +646,7 @@ def _make_pass_instance(
pass_type_list = [
PASS_TYPE_QUALIFIERS[q] for q in qualifier_ids if q in PASS_TYPE_QUALIFIERS
]
pass_type = pass_type_list[0] if len(pass_type_list) > 0 else "not_specified"
pass_type = pass_type_list[0] if len(pass_type_list) > 0 else "unspecified"
pass_type_hiearchy = ["cross", "through_ball", "long_ball"]
for pass_type_option in pass_type_hiearchy:
if pass_type_option in pass_type_list:
Expand Down Expand Up @@ -686,19 +693,23 @@ def _make_pass_instance(
datetime=_get_valid_opta_datetime(event),
start_x=x_start,
start_y=y_start,
pitch_size=pitch_dimensions,
team_side="home" if int(event.attrs["team_id"]) != away_team_id else "away",
_xt=-1.0,
outcome=outcome,
team_id=int(event.attrs["team_id"]),
team_side="home" if int(event.attrs["team_id"]) != away_team_id else "away",
pitch_size=pitch_dimensions,
player_id=int(event.attrs["player_id"]),
jersey=players.loc[
players["id"] == int(event.attrs["player_id"]), "shirt_num"
].iloc[0],
outcome=bool(event.attrs["outcome"]),
related_event_id=MISSING_INT,
body_part="unspecified",
possession_type="unspecified",
set_piece=set_piece,
_xt=-1.0,
outcome_str=outcome,
end_x=x_end,
end_y=y_end,
pass_type=pass_type,
set_piece=set_piece,
)


Expand Down Expand Up @@ -758,9 +769,7 @@ def _make_shot_event_instance(
for q in qualifiers
if int(q["qualifier_id"]) in SHOT_ORIGINS_QUALIFIERS
]
type_of_play = (
type_of_play_list[0] if len(type_of_play_list) > 0 else "regular_play"
)
type_of_play = type_of_play_list[0] if len(type_of_play_list) > 0 else "open_play"

body_part_list = [
BODY_PART_QUALIFIERS[int(q["qualifier_id"])]
Expand All @@ -778,53 +787,50 @@ def _make_shot_event_instance(
first_touch = (
event.find("Q", {"qualifier_id": str(FIRST_TOUCH_QUALIFIER)}) is not None
)

created_oppertunity_list = [
int(q["qualifier_id"])
for q in qualifiers
if int(q["qualifier_id"]) in CREATED_OPPERTUNITY_QUALIFIERS
]
created_oppertunity = (
CREATED_OPPERTUNITY_QUALIFIERS[created_oppertunity_list[0]]
if len(created_oppertunity_list) > 0
else "regular_play"
)

if event.find("Q", {"qualifier_id": str(RELATED_EVENT_QUALIFIER)}) is not None:
related_event_id = int(
event.find("Q", {"qualifier_id": str(RELATED_EVENT_QUALIFIER)})["value"]
)
else:
related_event_id = MISSING_INT

# set piece
set_piece_list = [
SET_PIECE_QUALIFIERS[int(q["qualifier_id"])]
for q in qualifiers
if int(q["qualifier_id"]) in SET_PIECE_QUALIFIERS
]
set_piece = set_piece_list[0] if len(set_piece_list) > 0 else "no_set_piece"

if int(event.attrs["team_id"]) == away_team_id:
x_start *= -1
y_start *= -1

return ShotEvent(
player_id=int(event.attrs["player_id"]),
jersey=players.loc[
players["id"] == int(event.attrs["player_id"]), "shirt_num"
].iloc[0],
event_id=int(event.attrs["id"]),
period_id=int(event.attrs["period_id"]),
minutes=int(event.attrs["min"]),
seconds=int(event.attrs["sec"]),
datetime=_get_valid_opta_datetime(event),
start_x=x_start,
start_y=y_start,
_xt=-1.0,
pitch_size=pitch_dimensions,
team_side="home" if int(event.attrs["team_id"]) != away_team_id else "away",
team_id=int(event.attrs["team_id"]),
shot_outcome=shot_outcome,
pitch_size=pitch_dimensions,
player_id=int(event.attrs["player_id"]),
jersey=players.loc[
players["id"] == int(event.attrs["player_id"]), "shirt_num"
].iloc[0],
outcome=shot_outcome == "goal",
related_event_id=related_event_id,
body_part=body_part,
possession_type=type_of_play,
set_piece=set_piece,
_xt=-1.0,
outcome_str=shot_outcome,
y_target=y_target,
z_target=z_target,
body_part=body_part,
type_of_play=type_of_play,
first_touch=first_touch,
created_oppertunity=created_oppertunity,
related_event_id=related_event_id,
)


Expand Down Expand Up @@ -873,25 +879,28 @@ def _make_dribble_event_instance(
y_start *= -1

dribble_event = DribbleEvent(
player_id=int(event.attrs["player_id"]),
jersey=players.loc[
players["id"] == int(event.attrs["player_id"]), "shirt_num"
].iloc[0],
event_id=int(event.attrs["id"]),
period_id=int(event.attrs["period_id"]),
minutes=int(event.attrs["min"]),
seconds=int(event.attrs["sec"]),
datetime=_get_valid_opta_datetime(event),
start_x=x_start,
start_y=y_start,
pitch_size=pitch_dimensions,
_xt=-1.0,
team_side="home" if int(event.attrs["team_id"]) != away_team_id else "away",
team_id=int(event.attrs["team_id"]),
team_side="home" if int(event.attrs["team_id"]) != away_team_id else "away",
pitch_size=pitch_dimensions,
player_id=int(event.attrs["player_id"]),
jersey=players.loc[
players["id"] == int(event.attrs["player_id"]), "shirt_num"
].iloc[0],
outcome=bool(event.attrs["outcome"]),
related_event_id=related_event_id,
body_part="foot",
possession_type="unspecified",
set_piece="no_set_piece",
_xt=-1.0,
duel_type=duel_type,
outcome=bool(event.attrs["outcome"]),
has_opponent=True, # opta take ons are always against an opponent
with_opponent=True, # opta take ons are always against an opponent
)

return dribble_event
Expand Down Expand Up @@ -987,6 +996,6 @@ def _update_pass_outcome(
else:
related_pass = related_pass.iloc[0]
# assign assist to pass
pass_events[related_pass["event_id"]].outcome = "assist"
pass_events[related_pass["event_id"]].outcome_str = "assist"

return pass_events
Loading

0 comments on commit 6103cdf

Please sign in to comment.