pickle serialization: module ‘__main__’ has no attribute ‘tokenize’… here is a solution to the problem.
pickle serialization: module ‘__main__’ has no attribute ‘tokenize’
I have a script that trains a machine learning model and saves it via pickle :
When I try to load the model to the website, I get the error message: “Module ‘main‘ does not have the property ‘tokenize'”.
I tried importing the function “tokenize” and copying it directly into the load script, but nothing worked
Full training script:
import sys
import pandas as pd
import numpy as np
from sqlalchemy import create_engine
import nltk
from nltk.tokenize import word_tokenize
from nltk.corpus import stopwords
import re
import seaborn as sns
from nltk.stem.wordnet import WordNetLemmatizer
from nltk.stem.porter import PorterStemmer
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.pipeline import Pipeline
from sklearn.feature_extraction.text import CountVectorizer
from sklearn.feature_extraction.text import TfidfTransformer
from sklearn.multioutput import MultiOutputClassifier
from sklearn.ensemble import RandomForestClassifier
from sklearn.ensemble import AdaBoostClassifier
from sklearn.model_selection import train_test_split
from sklearn.metrics import classification_report, fbeta_score, make_scorer
from sklearn.model_selection import GridSearchCV
from sklearn.externals import joblib
import pickle
nltk.download('punkt')
nltk.download('stopwords')
nltk.download('wordnet')
def load_data(database_filepath):
"""takes path to db as input and loads data. Return X, Y and target_names"""
engine = create_engine('sqlite:///{}'.format(database_filepath))
df = pd.read_sql('disaster_data', engine)
X = df.message.values
Y = df.drop(['message', 'id', 'original', 'genre'], axis=1).values
target_names = df.drop(['message', 'id', 'original', 'genre'], axis=1).columns
return X, Y, target_names
def tokenize(text):
"""Takes a text as input an returns a list of tokenized words"""
stop_words = stopwords.words("english")
text = re.sub(r"[^a-zA-Z0-9]", " ", text).lower().strip()
words = word_tokenize(text)
clean_words = [w for w in words if w not in stopwords.words("english")]
tokens = [WordNetLemmatizer().lemmatize(w) for w in words if w not in stop_words]
clean_tokens = [PorterStemmer().stem(w) for w in tokens]
return clean_tokens
def build_model():
"""Builds a model. returns a GridSearchCV object"""
pipeline = Pipeline([
('vect', CountVectorizer(tokenizer=tokenize)),
('tfidf', TfidfTransformer()),
('clf', MultiOutputClassifier(RandomForestClassifier(), n_jobs=1)),
])
parameters = {'clf__estimator__max_depth': [30],
'clf__estimator__min_samples_leaf': [5],
'clf__estimator__min_samples_split': [5],
'clf__estimator__n_estimators': [100]}
return GridSearchCV(estimator=pipeline, param_grid=parameters, verbose=10, n_jobs=1)
def evaluate_model(model, X_test, Y_test, category_names):
"""Takes model, X_test, Y_test and category names as input and evaluates model"""
y_pred = model.predict(X_test)
print("Accuracy of the model :", (y_pred == Y_test).mean())
for i in y_pred:
print(classification_report(Y_test, y_pred, target_names=category_names))
break
def save_model(model, model_filepath):
"""Takes model and path for saving as input and saves the model"""
pickle.dump(model, open(model_filepath, 'wb'))
# Uncommetn for joblib saving
# joblib.dump(model, model_filepath)
def main():
"""Main function"""
if len(sys.argv) == 3:
database_filepath, model_filepath = sys.argv[1:]
print('Loading data...\n DATABASE: {}'.format(database_filepath))
X, Y, category_names = load_data(database_filepath)
X_train, X_test, Y_train, Y_test = train_test_split(X, Y, test_size=0.2)
print('Building model...')
model = build_model()
print('Training model...')
model.fit(X_train, Y_train)
print('Evaluating model...')
evaluate_model(model, X_test, Y_test, category_names)
print('Saving model...\n MODEL: {}'.format(model_filepath))
save_model(model, model_filepath)
print('Trained model saved!')
else:
print('Please provide the filepath of the disaster messages database '\
'as the first argument and the filepath of the pickle file to '\
'save the model to as the second argument. \n\nExample: python '\
'train_classifier.py .. /data/DisasterResponse.db classifier.pkl')
if __name__ == '__main__':
main()
Load script:
import json
import plotly
import pandas as pd
import nltk
import pickle
from nltk.stem import WordNetLemmatizer
from flask import Flask
from flask import render_template, request, jsonify
from plotly.graph_objs import Bar
from sklearn.externals import joblib
from sklearn.feature_extraction.text import CountVectorizer
from sqlalchemy import create_engine
from nltk.corpus import stopwords
from flask import render_template
from wrangling_scripts.wrangle_data import return_figures
from nltk.stem.wordnet import WordNetLemmatizer
from nltk.stem.porter import PorterStemmer
from train_classifier_for_web import tokenize
nltk.download('punkt')
nltk.download('stopwords')
nltk.download('wordnet')
app = Flask(__name__)
def tokenize(text):
"""Takes a text as input an returns a list of tokenized words"""
stop_words = stopwords.words("english")
text = re.sub(r"[^a-zA-Z0-9]", " ", text).lower().strip()
words = word_tokenize(text)
clean_words = [w for w in words if w not in stopwords.words("english")]
tokens = [WordNetLemmatizer().lemmatize(w) for w in words if w not in stop_words]
return [PorterStemmer().stem(w) for w in tokens]
return clean_tokens
@app.before_first_request
def main():
try:
engine = create_engine('sqlite:///DisasterResponse.db')
df = pd.read_sql_table('disaster_data', engine)
except:
print("path error to sql db")
try:
model = joblib.load('web_model.sav','rb')
except Exception as e:
print("cant load model", e)
Solution
Resolved:
I tried to keep the tokenized function in a separate module and import it the same way in the training script and loading script