prediction.py

import streamlit as st
import pandas as pd 
import numpy as np
import seaborn as sns
import json 
import transformers

# Load model
with open('list_num_columns.txt', 'r') as file_1:
    list_num_cols = json.load(file_1)

with open('list_cat_columns.txt', 'r') as file_2:
    list_cat_cols = json.load(file_2)

with open('model_scaler.pkl', 'rb') as file_3:
    model_scaler = pickle.load(file_3)

with open('model_encoder.pkl', 'rb') as file_4:
    model_encoder = pickle.load(file_4)

with open('model_lin_reg.pkl', 'rb') as file_5:
    model_lin_reg = pickle.load(file_5)

with st.form('form_fifa_2022'):
    name = st.text_input('Name', value='', help='Player name')
    age = st.number_input('Age', 
                        min_value=12, 
                        max_value=48, 
                        value=28, 
                        step=2,
                        help='Player age')
    weight = st.number_input('Weight',
                            min_value=30,
                            max_value=100,
                            value=80,
                            help='Player weight')
    height = st.number_input('Height',
                            min_value=140,
                            max_value=210,
                            value=180,
                            help='Player height')
    price = st.slider('Price', 0, 200000000, 0)
    st.markdown('---')
    defense = st.radio('Defending Work Rate',
                    ('Low','Medium','High'),
                    index=1)
    attack = st.radio('Attacking Work Rate',
                    ('Low','Medium','High'),
                    index=1)
    st.markdown('---')
    pace = st.number_input('Pace', 0, 100, 50)
    shoot = st.number_input('Shooting', 0, 100, 50)
    passing = st.number_input('Passing', 0, 100, 50)
    dribble = st.number_input('Dribbling', 0, 100, 50)
    defend = st.number_input('Defending', 0, 100, 50)
    physical = st.number_input('Phisicality', 0, 100, 50)
    st.markdown('---')
    submitted = st.form_submit_button('Predict')


data_inf = {
    'Name': name,
    'Age': age,
    'Weight': weight,
    'Height': height,
    'Price': price,
    'AttackRate': attack,
    'DefenseRate': defense,
    'PaceTotal': pace,
    'ShootingTotal': shoot,
    'PassingTotal': passing,
    'DribblingTotal': dribble,
    'DefendingTotal': defend,
    'PhysicalityTotal': physical
}
data_inf = pd.DataFrame([data_inf])
st.dataframe(data_inf)

if submitted:
    # Numeric-Categoric split
    data_inf_num = data_inf[list_num_cols]
    data_inf_cat = data_inf[list_cat_cols]
    # Numeric
    data_inf_num_scaled = model_scaler.transform(data_inf_num)
    # Categoric
    data_inf_cat_encoded = model_encoder.transform(data_inf_cat)
    # Concatenate
    data_inf_final = np.concatenate([data_inf_num_scaled,data_inf_cat_encoded], axis=1)
    # Predict
    y_inf_pred = model_lin_reg.predict(data_inf_final)
    # Show prediction
    st.write('# Rating: ', str(int(y_inf_pred)))