GaussianNB对象无var_属性问题求助：Flask项目提交时报错

开发Python Flask项目时，编译正常，但提交表单后触发错误。使用的库包括numpy、flask、pickle、Pandas、Matplotlib。

错误信息 #

Traceback (most recent call last):
  File "C:\Users\sumit\AppData\Local\Programs\Python\Python310\lib\site-packages\flask\app.py", line 2091, in __call__
    return self.wsgi_app(environ, start_response)
  File "C:\Users\sumit\AppData\Local\Programs\Python\Python310\lib\site-packages\flask\app.py", line 2076, in wsgi_app
    response = self.handle_exception(e)
  File "C:\Users\sumit\AppData\Local\Programs\Python\Python310\lib\site-packages\flask\app.py", line 2073, in wsgi_app
    response = self.full_dispatch_request()
  File "C:\Users\sumit\AppData\Local\Programs\Python\Python310\lib\site-packages\flask\app.py", line 1519, in full_dispatch_request
    rv = self.handle_user_exception(e)
  File "C:\Users\sumit\AppData\Local\Programs\Python\Python310\lib\site-packages\flask\app.py", line 1517, in full_dispatch_request
    rv = self.dispatch_request()
  File "C:\Users\sumit\AppData\Local\Programs\Python\Python310\lib\site-packages\flask\app.py", line 1503, in dispatch_request
    return self.ensure_sync(self.view_functions[rule.endpoint])(**req.view_args)
  File "f:\Projects\Heart Disease Predictor\HeartDiseasesPredictor-master\main.py", line 31, in hello_world
    infProb = clf.predict_proba([enter])
  File "C:\Users\sumit\AppData\Local\Programs\Python\Python310\lib\site-packages\sklearn\naive_bayes.py", line 123, in predict_proba
    return np.exp(self.predict_log_proba(X))
  File "C:\Users\sumit\AppData\Local\Programs\Python\Python310\lib\site-packages\sklearn\naive_bayes.py", line 102, in predict_log_proba
    jll = self._joint_log_likelihood(X)
  File "C:\Users\sumit\AppData\Local\Programs\Python\Python310\lib\site-packages\sklearn\naive_bayes.py", line 487, in _joint_log_likelihood      
    n_ij = -0.5 * np.sum(np.log(2.0 * np.pi * self.var_[i, :]))
AttributeError: 'GaussianNB' object has no attribute 'var_'

相关代码 #

Main.py代码 #

from flask import Flask, render_template, request
import pickle
app = Flask(__name__)

file = open('model.pkl','rb')
clf = pickle.load(file)
file.close()

@app.route('/', methods = ["GET","POST"])
def hello_world():
    if request.method == "POST":
        myDict = request.form
        print(myDict)
        name = myDict['name']
        cp = int(myDict['cp'])
        trestbps = int(myDict['trestbps'])
        sex = int(myDict['sex'])
        age = int(myDict['age'])
        fbs= int(myDict['fbs'])
        chol = int(myDict['chol'])
        restecg = int(myDict['restecg'])
        slope = int(myDict['slope'])
        ca = int(myDict['ca'])
        thalach = int(myDict['thalach'])
        exang = int(myDict['exang'])
        oldpeak = float(myDict['oldpeak'])
        thal = int(myDict['thal'])
        print(myDict)
        enter = [age,sex,cp,trestbps,chol,fbs,restecg,thalach,exang,oldpeak,slope,ca,thal]
        print(enter)
        infProb = clf.predict_proba([enter])
        x = infProb[0][1]*100
        return render_template('show.html',uname=name,inf = int(x))
    return render_template('index.html')

if __name__ == '__main__':
    app.run(debug=True)

my_training.py代码 #

import numpy as np
import matplotlib.pyplot as plt
import pandas as pd
import pickle

dataset = pd.read_csv('heart.csv')
X = dataset.iloc[:, :-1].values
y = dataset.iloc[:, -1].values

from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.25, random_state = 0)

#Naive Bayes Classifier
from sklearn.naive_bayes import GaussianNB
classifier = GaussianNB()
classifier.fit(X_train, y_train)

infProb = classifier.predict_proba([[58,1,2,140,211,1,0,165,0,0,2,0,2]])
print(infProb[0][1]*100)

#creating the file to save the model
file = open('model.pkl', 'wb')
pickle.dump(classifier, file)
file.close()

解决方案 #

这个错误核心原因是训练模型和加载模型时的scikit-learn版本不一致。var_是GaussianNB在新版本scikit-learn中训练后生成的属性，若训练用旧版本、加载用新版本，就会出现属性缺失问题。

解决步骤：

1. 统一环境版本：
   • 查看Flask运行环境的scikit-learn版本：pip show scikit-learn
   • 在训练环境安装相同版本：pip install scikit-learn==x.x.x（替换x.x.x为查到的版本）
2. 重新运行my_training.py生成新的model.pkl
3. 重启Flask服务测试

另外，推荐用joblib替代pickle保存scikit-learn模型，对这类模型的序列化支持更友好：

• 修改训练代码的保存部分：

  import joblib
  joblib.dump(classifier, 'model.joblib')
  

• 修改Main.py的加载部分：

  import joblib
  clf = joblib.load('model.joblib')
  

内容的提问来源于stack exchange，提问作者Sumit