无论输入值如何，序数编码都会显示相同的值，从而使预测结果相同

Question

我正在建立一个模型，可以预测患者被诊断患有慢性肾脏病 以下是我使用的步骤；

• 导入训练模型和数据集所需的库
• 用随机值插补数值列，用众数插补分类列
• 对分类列执行原始编码
• 将数据集拆分为训练值和测试值
• 数据集中存在偏差值，因此我使用 SMOTE 库生成一些虚拟记录以使数据平衡
• 我使用决策树分类器作为模型

这是我的代码；

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
import math
import warnings
warnings.filterwarnings(action='ignore')


df = pd.read_csv('kidney_disease.csv')
cat_cols = [col for col in df.columns if df[col].dtype == 'object']
num_cols = [col for col in df.columns if df[col].dtype != 'object']

from sklearn.preprocessing import OrdinalEncoder
enc = OrdinalEncoder()

df_copied = df.copy()
X_enc = enc.fit_transform(df_copied[cat_cols])
df_copied[cat_cols] = pd.DataFrame(X_enc, columns=cat_cols)

x = df_copied.drop(['classification'], axis=1)
y = df_copied['classification']

from sklearn.model_selection import train_test_split

X_train, X_test, y_train, y_test = train_test_split(x, y, test_size = 0.3, random_state = 42)

from imblearn.over_sampling import SMOTE

smote = SMOTE(random_state=42)
X_resampled, y_resampled = smote.fit_resample(X_train, y_train)

dtc = grid_search_dtc.best_estimator_

# accuracy score, confusion matrix and classification report of decision tree

dtc_acc = accuracy_score(y_test, dtc.predict(X_test))

dtcClassifier = 'dtc_model.pkl'
with open(dtcClassifier, 'wb') as file:
    pickle.dump(dtc, file)

我使用streamlit库来实现我的模型

with open('dtc_model.pkl', 'rb') as file:  
    model = pickle.load(file)

# Define the column names
cols = ["age", "bp", "sg", "al", "su", "rbc", "pc", "pcc", "ba", "bgr", "bu", "sc", "sod", "pot", "hemo", "pcv", "wc", "rc", "htn", "dm", "cad", "appet", "pe", "ane"]

def main():
    st.title("Kidney Disease Prediction Using Hybrid Model")
  
   
        
    html_temp = """
    <div style="background:#025246 ;padding:10px">
    <h2 style="color:white;text-align:center;">KD Prediction App </h2>
    </div>
    """
    st.markdown(html_temp, unsafe_allow_html = True)
    
    # Define input fields
    age = st.text_input("Age", 0)
    bp = st.text_input("Blood Pressure", 0)
    sg = st.selectbox("Specific Gravity", [0, 1.005, 1.010, 1.015, 1.020, 1.025])
    al = st.selectbox("Albumin", [0, 1, 2, 3, 4, 5])
    su = st.selectbox("Sugar", [0, 1, 2, 3, 4, 5])
    rbc = st.selectbox("Red Blood Cells", ["", "normal", "abnormal"])
    pc = st.selectbox("Pus Cell", ["", "normal", "abnormal"])
    pcc = st.selectbox("Pus Cell clumps", ["", "present", "notpresent"])
    ba = st.selectbox("Bacteria", ["", "present", "notpresent"])
    bgr = st.text_input("Blood Glucose Random", 0)
    bu = st.text_input("Blood Urea", 0)
    sc = st.text_input("Serum Creatinine", 0)
    sod = st.text_input("Sodium", 0)
    pot = st.text_input("Potassium", 0)
    hemo = st.text_input("Hemoglobin", 0)
    pcv = st.text_input("Packed Cell Volume", 0)
    wc = st.text_input("White Blood Cell Count", 0)
    rc = st.text_input("Red Blood Cell Count", 0)
    htn = st.selectbox("Hypertension", ["", "yes", "no"])
    dm = st.selectbox("Diabetes Mellitus", ["", "yes", "no"])
    cad = st.selectbox("Coronary Artery Disease", ["", "yes", "no"])
    appet = st.selectbox("Appetite", ["", "good", "poor"])
    pe = st.selectbox("Pedal Edema", ["", "yes", "no"])
    ane = st.selectbox("Anemia", ["", "yes", "no"])
    
   
    
    if st.button("Predict"):
        
        # Convert data to DataFrame
        data = {
                'age': int(age), 
                'bp': float(bp), 
                'sg': sg, 
                'al': al, 
                'su': su, 
                'rbc': rbc, 
                'pc': pc,
                'pcc': pcc,
                'ba': ba, 
                'bgr': float(bgr), 
                'bu': float(bu), 
                'sc': float(sc), 
                'sod': float(sod), 
                'pot': float(pot), 
                'hemo': float(hemo), 
                'pcv': float(pcv), 
                'wc': float(wc), 
                'rc': float(rc), 
                'htn': htn, 
                'dm': dm, 
                'cad': cad, 
                'appet': appet, 
                'pe': pe, 
                'ane': ane
        }
        
        df = pd.DataFrame([data], columns=cols)
        # Check if all input fields are filled
        # Convert data to DataFrame
       
        
        
        cat_cols = [col for col in df.columns if df[col].dtype == 'object']
        enc = OrdinalEncoder()

        df_copied = df.copy()
        df_copied[cat_cols] = enc.fit_transform(df_copied[cat_cols])
        
        print(pd.DataFrame(df_copied))
        
        prediction = model.predict(df_copied)
        print(prediction[0])
        
        # Display prediction result
        if prediction[0] == 1:
                st.write("Positive")
        else:
                st.write("Negative")



if __name__ == '__main__':
    main()

当我以 Dataframe 格式打印 df_copied 时，我注意到 cat_cols 始终为 0

我尝试了 LabelEncoder() 但是一样的

Answer 1

您需要存储训练中的序数编码器，并使用它来转换来自 Streamlit 的传入数据，而不是创建一个新的编码器（它只能看到一个值，因此只能为其分配值 0，无论该值是什么） ).