Added example: SVM-based text sentiment classifier using scikit-learn

svm_sentiment.py

@@ -0,0 +1,42 @@
+"""
+svm_sentiment.py
+SVM-based Sentiment Analysis Example for TextBlob contribution.
+
+-----------------
+Example sentiment classifier using Support Vector Machines (SVM)
+"""
+
+from sklearn.feature_extraction.text import TfidfVectorizer
+from sklearn.pipeline import make_pipeline
+from sklearn.svm import LinearSVC
+
+
+class SVMSentimentClassifier:
+    def __init__(self):
+        self.model = make_pipeline(TfidfVectorizer(), LinearSVC())
+
+    def train(self, texts, labels):
+        """Train the classifier with labeled tetx data"""
+        self.model.fit(texts, labels)
+
+    def predict(self, text):
+        return self.model.predict([text])[0]
+
+
+if __name__ == "__main__":
+    texts = [
+        "I love this movie!",
+        "That was terrible.",
+        "It's okay, not bad",
+        "Absolutely amazing",
+        "Worst thing ever.",
+        "Fine but forgettable.",
+    ]
+    labels = ["positive", "negative", "neutral", "positive", "negative", "neutral"]
+
+    clf = SVMSentimentClassifier()
+    clf.train(texts, labels)
+
+    test = ["This was awesome!", "I didn't like it.", "It's average."]
+    for t in test:
+        print(t, "->", clf.predict(t))

test_svm_sentiment.py

@@ -0,0 +1,45 @@
+"""
+test_svm_sentiment.py
+unit tests for SVMSentimentClassifier.
+"""
+
+import pytest
+from svm_sentiment import SVMSentimentClassifier
+
+
+@pytest.fixture
+def trained_classifier():
+    texts = [
+        "I love this movie!",
+        "That was terrible.",
+        "It's okay, not bad",
+        "Absolutely amazing",
+        "Worst thing ever.",
+        "Fine but forgettable.",
+    ]
+    labels = ["positive", "negative", "neutral", "positive", "negative", "neutral"]
+
+    clf = SVMSentimentClassifier()
+    clf.train(texts, labels)
+    return clf
+
+
+def test_positive_sentiment(trained_classifier):
+    result = trained_classifier.predict("This is fantastic!")
+    assert result in ["positive", "negative", "neutral"]
+
+
+def test_negative_sentiment(trained_classifier):
+    result = trained_classifier.predict("I really hated it")
+    assert result in ["positive", "negative", "neutral"]
+
+
+def test_neutral_sentiment(trained_classifier):
+    result = trained_classifier.predict("It was fine, nothing special.")
+    assert result in ["positive", "negative", "neutral"]
+
+
+def test_model_training_required():
+    clf = SVMSentimentClassifier()
+    with pytest.raises(Exception):
+        clf.predict("I love it!")