app.py

import streamlit as st
import pickle
import pandas as pd
import joblib

loaded_model = pickle.load(open('trained_model.sav', 'rb'))
preprocessor = joblib.load("preprocessor.pkl")

def obj_to_int(data):
    # Mengganti nilai pada kolom planting_period
    one_hot_enc = pd.get_dummies(data, columns=['planting_period', 'variety'])
    
    return one_hot_enc
    
def transform_normalize(data, normalizer):
    normalize_data = pd.DataFrame(normalizer.transform(data))
    normalize_data.columns = data.columns
    normalize_data.index = data.index
    return normalize_data
   
def preprocess(data):
    # 1. Encoding Categorical Value
    X_encoded = obj_to_int(data)

    required_features = ['planting_period_MT1',
                        'planting_period_MT2',
                        'planting_period_MT3',
                        'variety_Umum',
                        'variety_Hibrida'
                        ]

    existing_features = set(X_encoded.columns)

    # Ambil perbedaan antara fitur yang dibutuhkan dan yang sudah ada
    missing_features = set(required_features) - existing_features

    list_missing_features = list(missing_features)

    for feature in list_missing_features:
        X_encoded[feature] = 0

    # urutkan column sesuai data train
    column_order = ['completion_rate',
                    'complete_time',
                    'profit','harvest',
                    'planting_period_MT1',
                    'planting_period_MT2',
                    'planting_period_MT3',
                    'variety_Hibrida',
                    'variety_Umum'
                    ]

    X_encoded = X_encoded[column_order]

    # 2. Normalization
    X_clean = transform_normalize(data = X_encoded,
                                    normalizer = preprocessor['normalizer'])

    return X_clean

def calculate_harvest_score(row):
    variety = row['variety']
    harvest = row['harvest']
    variety_umum = ['Umum', 'Sintanur', 'Mentik']
    variety_hibrida = ['Hibrida', 'Mapan']

    if variety in variety_umum:
        if harvest < 3000:
            score_harvest = 46.79
        elif 3000 <= harvest <= 5000:
            score_harvest = 64.24
        elif 5001 <= harvest <= 7000:
            score_harvest = 93.57
        else:
            score_harvest = 102.83
            
    elif variety in variety_hibrida:
        if harvest < 5000:
            score_harvest = 40.55
        elif 5001 <= harvest <= 7000:
            score_harvest = 54.02
        elif 7001 <= harvest <= 9000:
            score_harvest = 68.34
        else:
            score_harvest = 85.17
    else:
        score_harvest = 0
    
    return score_harvest

def calculate_profit_score(row):
    profit = row['profit']
    
    if profit <= 0:
        score_profit = 0
    elif 1 <= profit <= 2000000:
        score_profit = 27.84
    else:
        score_profit = 39.5
    
    return score_profit

def calculate_complete_time_score(row):
    complete_time = row['complete_time']

    if complete_time <= 3:
        score_complete_time = 10.21
        return score_complete_time
    
    elif 4 <= complete_time <= 7:
        score_complete_time = 6.69
        return score_complete_time
    
    else:
        score_complete_time = 1
        return score_complete_time

def calculate_completion_rate_score(row):
    completion_rate = row['completion_rate']

    if completion_rate <= 60:
        score_completion_rate = 17.31
        return score_completion_rate
    
    elif 61 <= completion_rate <= 70:
        score_completion_rate = 21.04
        return score_completion_rate
    
    elif 71 <= completion_rate <= 80:
        score_completion_rate = 25.21
        return score_completion_rate
    
    elif 81 <= completion_rate <= 90:
        score_completion_rate = 30.1
        return score_completion_rate
    
    else:
        score_completion_rate = 34.4
        return score_completion_rate

def total_score(data):
    harvest_score = data.apply(calculate_harvest_score, axis=1)
    profit_score = data.apply(calculate_profit_score, axis=1)
    completion_rate_score = data.apply(calculate_completion_rate_score, axis=1)
    complete_time_score = data.apply(calculate_complete_time_score, axis=1)

    score = harvest_score + profit_score + complete_time_score + completion_rate_score

    return score

def map_class_to_label(class_score):
    class_labels = {
        0: "Sangat Baik",
        1: "Baik",
        2: "Cukup",
        3: "Buruk",
        4: "Sangat Buruk"
    }
    return class_labels.get(class_score, "Unknown")

def calculate_scoring(data):
    score = total_score(data)
    clean_data = preprocess(data)
    hasil = loaded_model.predict(clean_data)

    # Mapping nilai pada hasil ke label yang diinginkan
    hasil_labels = [map_class_to_label(pred_class) for pred_class in hasil]

    print('Hasil adalah', hasil_labels,
          '\nTotal Score: ', score)
    return hasil_labels, score

def main():
    st.title("Project Scoring")

    # Membuat tombol untuk upload file CSV
    uploaded_file = st.file_uploader("Upload CSV file", type=["csv"])

    if uploaded_file is not None:
        # Membaca file CSV menjadi DataFrame
        data = pd.read_csv(uploaded_file)

        # Menampilkan preview dari data yang diupload
        st.write("Preview Data:")
        st.write(data.head())

        # Membuat tombol "Calculate Scoring"
        if st.button("Calculate Scoring"):
            # Memanggil fungsi untuk menghitung scoring
            hasil_scoring = calculate_scoring(data)

            # Menampilkan hasil scoring
            st.write("Hasil Scoring:")
            hasil_df = pd.DataFrame({
                'Kategori': hasil_scoring[0],
                'Total Score': hasil_scoring[1]
            })
            st.table(hasil_df)

if __name__ == "__main__":
    main()