app.py

import gradio as gr
from joblib import load
import pandas as pd

path='./huggingface_final_df_deployment.csv'
df=pd.read_csv(path)
df['DATE']=pd.to_datetime(df['DATE'])

y_values = ['L', 'W']
y=pd.Series(y_values)
y=y.astype('category')

model=load('./model_LogisticRegression.joblib')
#aquí indicamos cuál es el valor de las clases
label_encoding=load('./label_classes.joblib') # aquí traemos el orden de clasificación de las categoría
idx_labels=label_encoding.transform(y.cat.categories) # reordeno mis categorías con esos índices
class_names=y.cat.categories[idx_labels] # aquí damos los valores de las categorías en el orden que teníamos guardados
def run_my_model(Home_Team, Away_Team): #aquí ponemos que es solo una entrada pero es UNA TABLA la entrada
    
    #Paso1: Función de HomeTeam
    def home_team_recent_record(Home_Team):
        filtered_df=df[df['TEAM_ABBREVIATION']==Home_Team]
        most_recent_record=filtered_df[filtered_df['DATE']==filtered_df['DATE'].max()]
    
        return most_recent_record

     #Paso2: Función de AwayTeam
    def away_team_recent_record(Away_Team):
        filtered_df=df[df['TEAM_ABBREVIATION']==Away_Team]
        most_recent_record=filtered_df[filtered_df['DATE']==filtered_df['DATE'].max()]
    
        return most_recent_record
    
    #Paso3: Dataframe con últimos records de HomeTeam
    df_home_team_latest_record=home_team_recent_record(Home_Team)
    df_home_team_latest_record=df_home_team_latest_record.reset_index()
    
    #Paso4: Dataframe con últimos records de AwayTeam
    df_away_team_latest_record=away_team_recent_record(Away_Team)
    df_away_team_latest_record=df_away_team_latest_record.reset_index()
    
    #Paso5: Cálculos
    HOME_AWAY="HOME"
    WINS_DIFFERENTIAL=df_home_team_latest_record.loc[0,"W"]-df_away_team_latest_record.loc[0,"W"]
    WINNING_PERCENTAGE_DIFFERENTIAL=(df_home_team_latest_record.loc[0,'W_PCT']-df_away_team_latest_record.loc[0,'W_PCT'])*100
    AWAY_WINS_DIFFERENTIAL=df_home_team_latest_record.loc[0,'AWAY_WINS']-df_away_team_latest_record.loc[0,'AWAY_WINS']
    WINS_IN_LAST_10_GAMES=df_home_team_latest_record.loc[0,'LAST_10_GAME_WINS']
    WINS_IN_LAST_5_GAMES=df_home_team_latest_record.loc[0,'LAST_5_GAME_WINS']
    BACK_TO_BACK="NO_B2B"
    
    #Paso6: Convierto a un input dataframe
    data = {
            'HOME/AWAY': [HOME_AWAY],
            'WINS_DIFFERENTIAL': [WINS_DIFFERENTIAL],
            'WINNING PERCENTAGE DIFFERENTIAL': [WINNING_PERCENTAGE_DIFFERENTIAL],
            'AWAY WINS DIFFERENTIAL': [AWAY_WINS_DIFFERENTIAL],
            'WINS IN LAST 10 GAMES': [WINS_IN_LAST_10_GAMES],
            'WINS IN LAST 5 GAMES': [WINS_IN_LAST_5_GAMES],
            'BACK-TO-BACK': [BACK_TO_BACK]}

    input_df = pd.DataFrame(data)
    
    #Paso7: Creación del If
    if class_names[model.predict(input_df)][0]=="W":
        Winning_Team=Home_Team
        Winning_Proba=model.predict_proba(input_df)[0][1]
    else:
        Winning_Team=Away_Team
        Winning_Proba=model.predict_proba(input_df)[0][0]
    
    return f"The team with higher probability of winning is: {Winning_Team} with a probability of {Winning_Proba * 100:.2f}%"

gui= gr.Interface(
    fn=run_my_model,
    inputs=[gr.Dropdown(['ATL', 'BKN', 'BOS', 'CHA', 'CHI', 'CLE', 'DAL', 'DEN', 'DET', 'GSW', 'HOU', 'IND', 'LAC', 'LAL', 'MEM', 'MIA', 'MIL', 'MIN', 'NOP', 'NYK', 'OKC', 'ORL', 'PHI', 'PHX', 'POR', 'SAC', 'SAS', 'TOR', 'UTA', 'WAS'],label="HOME TEAM"),gr.Dropdown(['ATL', 'BKN', 'BOS', 'CHA', 'CHI', 'CLE', 'DAL', 'DEN', 'DET', 'GSW', 'HOU', 'IND', 'LAC', 'LAL', 'MEM', 'MIA', 'MIL', 'MIN', 'NOP', 'NYK', 'OKC', 'ORL', 'PHI', 'PHX', 'POR', 'SAC', 'SAS', 'TOR', 'UTA', 'WAS'],label="AWAY TEAM")],
    outputs="text",
    live=False,
    title='NBA regular season games outcome prediction',
    description='IMPORTANT: The predictions generated by this AI tool are for informational purposes only and should not be considered as any type of advice. Past performance is not indicative of future results. This predictions are only for NBA regular season games.'
)

gui.launch()