transform into Prefect flow

main.py

@@ -11,47 +11,44 @@ from nltk.stem import WordNetLemmatizer
 from sklearn.feature_extraction.text import TfidfVectorizer
 from sklearn.linear_model import LogisticRegression
 from sklearn.model_selection import train_test_split
+from prefect import task, Flow
+from pandas import DataFrame
+from typing import List
 
 nltk.download('stopwords')
 nltk.download('wordnet')
 nltk.download('punkt')
 
-train = pd.read_csv('train.csv')
-test = pd.read_csv('test.csv')
-
-train.head()
-train_count=train.count()
-print(train_count)
-print(train_count/train_count[0]*100)
-train = train.drop(['keyword', 'location'], axis = 1)
-
-test.head()
-test.describe()
 
 # remove urls, handles, and the hashtag from hashtags (taken from https://stackoverflow.com/questions/8376691/how-to-remove-hashtag-user-link-of-a-tweet-using-regular-expression)
 def remove_urls(text):
     new_text = ' '.join(re.sub("(@[A-Za-z0-9]+)|([^0-9A-Za-z \t])|(\w+:\/\/\S+)"," ",text).split())
     return new_text
 
+
 # make all text lowercase
 def text_lowercase(text):
     return text.lower()
 
+
 # remove numbers
 def remove_numbers(text):
     result = re.sub(r'\d+', '', text)
     return result
 
+
 # remove punctuation
 def remove_punctuation(text):
     translator = str.maketrans('', '', string.punctuation)
     return text.translate(translator)
 
+
 # tokenize
 def tokenize(text):
     text = word_tokenize(text)
     return text
 
+
 # remove stopwords
 stop_words = set(stopwords.words('english'))
 def remove_stopwords(text):
@@ -64,6 +61,7 @@ def lemmatize(text):
     text = [lemmatizer.lemmatize(token) for token in text]
     return text
 
+
 def preprocessing(text):
     text = text_lowercase(text)
     text = remove_urls(text)
@@ -76,41 +74,98 @@ def preprocessing(text):
     return text
 
 
-pp_text_train = [] # our preprocessed text column
+@task
+def get_train_set() -> DataFrame:
+    train = pd.read_csv('train.csv')
+    train = train.drop(['keyword', 'location'], axis=1)
+    return train
+
+
+@task
+def get_test_set() -> DataFrame:
+    return pd.read_csv('test.csv')
+
+
+@task
+def preprocess_train(train: DataFrame) -> DataFrame:
+    pp_text_train = []
     for text_data in train['text']:
         pp_text_data = preprocessing(text_data)
         pp_text_train.append(pp_text_data)
-train['pp_text'] = pp_text_train # add the preprocessed text as a column
+    train['pp_text'] = pp_text_train
+    return train
 
-pp_text_test = [] # our preprocessed text column
+
+@task
+def preprocess_test(test: DataFrame) -> DataFrame:
+    pp_text_test = []
     for text_data in test['text']:
         pp_text_data = preprocessing(text_data)
         pp_text_test.append(pp_text_data)
-test['pp_text'] = pp_text_test # add the preprocessed text as a column
+    test['pp_text'] = pp_text_test
+    return test
 
-train_text_data = list(train['pp_text'])
-test_text_data = list(test['pp_text'])
+
+@task
+def prepare_vectorizer(train_data: DataFrame, test_data: DataFrame) -> TfidfVectorizer:
+    train_text_data = list(train_data['pp_text'])
+    test_text_data = list(test_data['pp_text'])
     corpus = train_text_data + test_text_data
-
     tf = TfidfVectorizer()
-# the vectorizer must be fit onto the entire corpus
     fitted_vectorizer = tf.fit(corpus)
+    return fitted_vectorizer
 
-# train
-train_transform = fitted_vectorizer.transform(train['pp_text'])
-y = train['target']
-# test
-test_transform = fitted_vectorizer.transform(test['pp_text'])
 
-X=train_transform
-X_train, X_test, y_train, y_test = train_test_split(X, y)
+@task
+def transform_train(vectorizer: TfidfVectorizer, train_set: DataFrame) -> DataFrame:
+    return vectorizer.transform(train_set)
 
+
+@task
+def transform_test(vectorizer: TfidfVectorizer, test_set: DataFrame) -> DataFrame:
+    return vectorizer.transform(test_set)
+
+
+@task
+def split_test_set(X: DataFrame, Y: DataFrame) -> dict:
+    X_train, X_test, y_train, y_test = train_test_split(X, Y)
+    return {'X_train': X_train, 'X_test': X_test, 'y_train': y_train, 'y_test': y_test}
+
+
+@task
+def train_model(X: DataFrame, Y: DataFrame) -> LogisticRegression:
     scikit_log_reg = LogisticRegression()
-model=scikit_log_reg.fit(X_train, y_train)
+    model = scikit_log_reg.fit(X, Y)
+    return model
 
-predictions = model.predict(X_test)
+
+@task
+def evaluate(model: LogisticRegression, X: DataFrame, Y: DataFrame) -> None:
+    predictions = model.predict(X)
     count = 0
-for guess, answer in zip(predictions, y_test):
+    for guess, answer in zip(predictions, Y):
         if guess == answer:
             count += 1
-print(count/len(y_test))
+    print("> model score: ", count/len(Y))
+
+
+if __name__ == "__main__":
+    with Flow("My First Flow!") as flow:
+        train_data = get_train_set()
+        test_data = get_test_set()
+
+        train_data = preprocess_train(train_data)
+        test_data = preprocess_test(test_data)
+
+        vectorizer = prepare_vectorizer(train_data, test_data)
+
+        vectorized_train_data = transform_train(vectorizer, train_data['pp_text'])
+        vectorized_test_data = transform_test(vectorizer, train_data['pp_text'])
+
+        splitted_data = split_test_set(vectorized_train_data, train_data['target'])
+        model = train_model(splitted_data['X_train'], splitted_data['y_train'])
+        evaluate(model, splitted_data['X_test'], splitted_data['y_test'])
+
+    flow.validate()
+    # flow.visualize()
+    flow.run()