Soccer analytics open-source demo

requirements.txt

kloppy
mplsoccer
streamlit
natsort

script.py

import os
from glob import glob

from natsort import natsorted

from mplsoccer.pitch import Pitch
from kloppy import load_opta_event_data, to_pandas
import streamlit as st

match_list = [os.path.basename(filename) for filename in glob("data/*")]

match = st.sidebar.selectbox("Select match", match_list, index=0)

# make sure you have a directory called "data" with directories containing the f24/f7 xml files
f24_filename = f"data/{match}/f24.xml"
f7_filename = f"data/{match}/f7.xml"

dataset = load_opta_event_data(
    f24_filename,
    f7_filename,
    options={"event_types": ["pass"]}
)
dataframe = to_pandas(dataset)

# This will create an unique Set of team/jersey no combos
players = {f"{event.team} - {event.player_jersey_no}" for event in dataset.events}

player = st.sidebar.selectbox("Select player", natsorted(players), index=0)
st.text(f"Selected match: {match} / Player: {player}")

team, jersey_no = player.split(" - ")

player_events = dataframe[(dataframe.team == team) & (dataframe.player_jersey_no == jersey_no)]

pitch = Pitch(pitch_color='#e7f1fa', line_zorder=1, line_color='black', pitch_type="opta")
fig, ax = pitch.draw()
pitch.kdeplot(
    player_events["position_x"],
    player_events["position_y"],
    ax=ax,
    shade=True,
    n_levels=50,
)

st.pyplot()

# run using: streamlit run script.py