app.py

import streamlit as st
import pandas as pd
import pickle
from pathlib import Path
import os

st.set_page_config(
    page_title="Player Recommender",
    page_icon=":soccer:"
)


@st.cache_data(show_spinner=False)
def getData():
    # loading outfield players' cleaned data and engine
    player_df = pd.read_pickle(r'data/outfield.pkl')
    with open(r'data/player_ID.pickle', 'rb') as file:
        player_ID = pickle.load(file)
    with open(r'data/engine.pickle', 'rb') as file:
        engine = pickle.load(file)

    # loading gk players' cleaned data and engine
    gk_df = pd.read_pickle(r'data/gk.pkl')
    with open(r'data/gk_ID.pickle', 'rb') as file:
        gk_ID = pickle.load(file)
    with open(r'data/gk_engine.pickle', 'rb') as file:
        gk_engine = pickle.load(file)

    return [player_df, player_ID, engine], [gk_df, gk_ID, gk_engine]


outfield_data, gk_data = getData()


header = st.container()
params = st.container()
result = st.container()
teams = st.container()
credit = st.container()


with header:
    st.title('Player Recommendation Tool ')

with params:
    st.text(' \n')
    st.text(' \n')
    st.header('Get recommendations')

    col1, col2, col3 = st.columns([1, 2.2, 0.8])
    with col1:
        radio = st.radio('Player type', ['Outfield players', 'Goal Keepers'])
    with col2:
        if radio == 'Outfield players':
            df, player_ID, engine = outfield_data
        else:
            df, player_ID, engine = gk_data
        players = sorted(list(player_ID.keys()))
        age_default = (min(df['Age']), max(df['Age']))
        query = st.selectbox('Player name', players,
                             help='Type without deleting a character. To search from a specific team, just type in the club\'s name.')
    with col3:
        foot = st.selectbox('Preferred foot', ['All', 'Right', 'Left'])

    col4, col5, col6, col7 = st.columns([0.7, 1, 1, 1])
    with col4:
        if radio == 'Outfield players':
            res, val, step = (5, 20), 10, 5
        else:
            res, val, step = (3, 10), 5, 1
        count = st.slider(
            'Number of results', min_value=res[0], max_value=res[1], value=val, step=step)
    with col5:
        comp = st.selectbox('League', ['All', 'Premier League', 'La Liga', 'Serie A', 'Bundesliga', 'Ligue 1'],
                            help='Leagues to get recommendations from. \'All\' leagues by default.')
    with col6:
        comparison = st.selectbox('Comparison with', ['All positions', 'Same position'],
                                  help='Whether to compare the selected player with all positions or just the same defined position in the dataset. \'All \
            positions\' by default.')
    with col7:
        age = st.slider('Age bracket', min_value=age_default[0], max_value=age_default[1], value=age_default,
                        help='Age range to get recommendations from. Drag the sliders on either side. \'All\' ages by default.')


with result:
    st.text(' \n')
    st.text(' \n')
    st.text(' \n')
    st.markdown('_showing recommendations for_ **{}**'.format(query))

    def getRecommendations(metric, df_type, league='All', foot='All', comparison='All positions', age=age_default, count=val):
        if df_type == 'outfield':
            df_res = df.iloc[:, [1, 3, 5, 6, 11, -1]].copy()
        else:
            df_res = df.iloc[:, [1, 3, 5, 6, 11]].copy()
        df_res['Player'] = list(player_ID.keys())
        df_res.insert(1, 'Similarity', metric)
        df_res = df_res.sort_values(by=['Similarity'], ascending=False)
        metric = [str(num) + '%' for num in df_res['Similarity']]
        df_res['Similarity'] = metric
        df_res = df_res.iloc[1:, :]

        if comparison == 'Same position' and df_type == 'outfield':
            q_pos = list(df[df['Player'] == query.split(' (')[0]].Pos)[0]
            df_res = df_res[df_res['Pos'] == q_pos]

        if league == 'All':
            pass
        else:
            df_res = df_res[df_res['Comp'] == league]

        if age == age_default:
            pass
        else:
            df_res = df_res[(df_res['Age'] >= age[0]) &
                            (df_res['Age'] <= age[1])]

        if foot == 'All' or df_type == 'gk':
            pass
        elif foot == 'Left':
            df_res = df_res[df_res['Foot'] == 'left']
        else:
            df_res = df_res[df_res['Foot'] == 'right']

        df_res = df_res.iloc[:count, :].reset_index(drop=True)
        df_res.index = df_res.index + 1
        if len(df) == 2040:
            mp90 = [str(round(num, 1)) for num in df_res['90s']]
            df_res['90s'] = mp90
        df_res.rename(columns={'Pos': 'Position',
                      'Comp': 'League'}, inplace=True)
        return df_res

    sims = engine[query]
    df_type = 'outfield' if len(df) == 2040 else 'gk'
    recoms = getRecommendations(sims, df_type=df_type, foot=foot,
                                league=comp, comparison=comparison, age=age, count=count)
    st.table(recoms)


with teams:
    st.text(' \n')
    st.header("Team")
    col1, col2, col3, col4 = st.columns([1, 1, 1, 1])

    with col1:
        st.image('images/bijay.jpg', width=150)
        st.write('**Bijay Sapkota**')

    with col2:
        st.image('images/ishan.jpg', width=150)
        st.write('**Ishan Panta**')

    with col3:
        st.image('images/manish.png', width=150)
        st.write('**Manish Shivabhakti**')

    with col4:
        st.image('images/priyanshu.jpg', width=150)
        st.write('**Priyanshu Sharma**')

with credit:
    st.text(' \n')
    st.text(' \n')

    @st.cache_data()
    def read_info(path):
        return Path(path).read_text(encoding='utf8')

    st.markdown(read_info('info.md'), unsafe_allow_html=True)