From 0bb8bd6f9a190610e29a96f7ee7fd5805f01185b Mon Sep 17 00:00:00 2001
From: Jane Heng <janeheng.ic@gmail.com>
Date: Sat, 11 Oct 2025 12:47:13 -0700
Subject: [PATCH] Add SVM-based sentiment analysis: source classification.

---
 .../svm_source_classifier.cpython-311.pyc     | Bin 0 -> 2512 bytes
 code/svm_source_classifier.py                 |  95 ++++++++++++++++++
 code/test_source_classifier.py                |  71 +++++++++++++
 3 files changed, 166 insertions(+)
 create mode 100644 code/__pycache__/svm_source_classifier.cpython-311.pyc
 create mode 100644 code/svm_source_classifier.py
 create mode 100644 code/test_source_classifier.py

diff --git a/code/__pycache__/svm_source_classifier.cpython-311.pyc b/code/__pycache__/svm_source_classifier.cpython-311.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..3fd6afcd7d7845c554465218f6b7ec1811f2b19e
GIT binary patch
literal 2512
zcmZuy&u<$=6rSB3ukDSK*ePvEX^17%mTigCrVW7>74?@&LYq=0O)C>=wb>opTd%$D
z>^iM28(9cJg#cAt5|EH04xuPP;@E!xiQ^jiV67A`oZw^>xb(!Eb?n-uqnWpF=e>P1
z^Sy80{xCQwA!y%z^H=7V5JG>jmrfJ*l-;jDSwlKXqA=1qoj180pX52F3#O36N$l&`
z4CKV5=<5M9m<uICzAl>KoRpM64}OA@19}K#L=S@;)FqHZWLO`_@@zF4OCvq<J(rB~
zhzybFCWp`#jCO+vgA4wUhswV%EL~S@t4KAXXr^l0M#>;GVH8UF6-l~KbgZ1}7@BID
zrFfd;2~{1U$5%@6OO~mwBqZquHPS|2HRCD6Byq!zo0jJLsE!nemT{bdeRgtE^ZhBY
ziAuFhB5he|lO!}NH>p!&l}w(RwaY7oJ2O^x@t*$HbV<#Ymhx9;t}JFQUz_^i`ssos
zEoLA*bTtl}o>)<B*hzcq@y^y^X~W9LQ^mZ-6zNZPWSR3uQpC|R%GAj`xo@YbRV>&r
zJG_`O^wbj39E%!dLSaf+Sh@s{!JL{UO2H@))5sH;2<C`G4b9fNFhbw}43^&A&q1sq
zr^lvq6{Pdq!Y0_aIut-vuFnU*(1)*Zce{~5_frZBpbFnJ&-cu8ce^X30t-kE%zFWc
zszzQGykLf?I-#}~Oj(x06!BgW6dP=!U0kt=<4G)IX~2Mhm6DvbN<0iMj*Kab>|tGX
zR8MqNng;hs`>EvB91BB-n2Y2usQ8Xy_N}kl>~mnj)Aui*U7k8Ky_~rZed6@8RUr9x
z?v@wH14p^E$Gvs}mo-Z#%l2yS1=K8*ys&Ahx*wk&g!p4HYz0IO?ZjSd4j*Xl8*2_v
zwt{#>{2PJzhl$$UzY=2R{atg6nPbcx6Kk_A2}KTk@$ULu<M71e()LGBN{x4KxHms+
z+|0NG85d`EVF{#l(*Z9c40giZPeJq(!0Dz0p-r%E!Du&!(Pppd+&15TJ2xTzR(Fjm
zU*R`;B6Yq4GN20A@5=!}Ao;?)Co<BijLu?~&^$3Ak8G<HGr^Ykgp}bhcHB|?ym`UW
zo)X9b^kye-^Z;`nXy?|q&^XMTfMK5l0f<LNzniQdI?*_E;-`ftA$Q_Jed0m`WaMHa
za<O)$iNl}Gemwgy{pfxjk2mnRi^rP-(b_foGJLPG9d%j|6d%n8psc|OH#(NSC!8Cu
z$Fq7?#l2YfqWkPxRQL+62-{r$b?lD=brAu6{JhN3eXyd8y<mY7-OwB_P@qQMp=^iV
z{{CX4P*Sj{BM0a>vx<Nyu-_8_P<Z;0yzpKeqT398xDfGp8`>eVs+vW(eo7nN*TH-W
zhW$2(8fr%OuMK`ZxSpy<k2Io3YO^~y{N=HRbnLPCbNJVYE8VI~w;CYvQUfo!o!-W_
zTSOpKm#GfC^fR@B`f*r>EmpZIUlppj8tBEg!d3Ww&Q--;yencw0GtDxkkzejW`UO?
z;L6QIsbJUDXAorvm4klmpzH$ESD84%#8D<#27E+q2D*<}2eVOH%s*ck+I52-2j4eg
z*cU+nIj7G*Dc7gx8q;%iJXO2;@b;tBdTPVmi5y%HZXB#f<VHlUUD?4SkEYj;ZXA95
zMjg*I@QjORo^jL*DM}8CsiMe2dBjo8Y@%S92K*K6M9#3aa+DG+VKWjWGM1JlrE=KT
z446?hGB3xxprYuOrYK(2&YEy9^9fcZq43)6>N?(;P7&29QldaXq^kb|OE^qtP<jI$
zFDQ%cJKPP|`A_+m+zakG`fPm;aUURdAXd8-#fSe#<kQ9H<I~VPz{8?R&eJonC}WV#
z{-G^_<G2<Makzz;cqs^C1aT)_G}+sm=$zZ%o9Gp{zc<lwx4-}2rxhCD46cQG<Y9q}
zwNO_?c<weh^hgOahprexIP$Rkn=tAMqs`cRuHS2OO+4n}vGu!kEH|+1c6uv-@FDQ@
Gq4*CR=5;gx

literal 0
HcmV?d00001

diff --git a/code/svm_source_classifier.py b/code/svm_source_classifier.py
new file mode 100644
index 0000000..2550989
--- /dev/null
+++ b/code/svm_source_classifier.py
@@ -0,0 +1,95 @@
+# -*- coding: utf-8 -*-
+"""
+SVM_source_classifier.py
+This code trains a SVM model to classify the source of a text document
+based on writing style and topic using the fetch_20newsgroups dataset.
+Removes headers, footers, and quotes to prevent data leakage.
+Author: Jane Heng
+Date: Oct.10.2025
+"""
+from sklearn.datasets import fetch_20newsgroups
+from sklearn.feature_extraction.text import TfidfVectorizer
+from sklearn.svm import SVC
+from sklearn.pipeline import make_pipeline
+from sklearn import metrics
+
+def load_data(categories):
+    """
+    Loads training and testing data from the 20 Newsgroups dataset for the specified categories.
+    Removes headers, footers, and quotes to reduce noise and prevent data leakage.
+
+    Args:
+        categories (list of str): List of category names to include.
+
+    Returns:
+        tuple: A tuple containing:
+            - train_data (list of str): Training text samples.
+            - train_target (list of int): Labels for training samples.
+            - test_data (list of str): Testing text samples.
+            - test_target (list of int): Labels for testing samples.
+            - target_names (list of str): Names of the target categories.
+    """
+    train_data = fetch_20newsgroups(
+        subset='train',
+        categories=categories,
+        remove=('headers', 'footers', 'quotes')
+    )
+    test_data = fetch_20newsgroups(
+        subset='test',
+        categories=categories,
+        remove=('headers', 'footers', 'quotes')
+    )
+    return train_data.data, train_data.target, test_data.data, test_data.target, train_data.target_names
+
+def train_model(X_train, y_train):
+    """
+    Trains a Support Vector Machine (SVM) classifier using TF-IDF features.
+
+    Args:
+        X_train (list of str): Training text data.
+        y_train (list of int): Corresponding labels for training data.
+
+    Returns:
+        sklearn.pipeline.Pipeline: A trained pipeline combining TF-IDF vectorization and SVM classification.
+    """
+    model = make_pipeline(TfidfVectorizer(), SVC(kernel='linear'))
+    model.fit(X_train, y_train)
+    return model
+
+def evaluate_model(model, X_test, y_test, target_names):
+    """
+    Evaluates the trained model on the test data and prints a classification report.
+
+    Args:
+        model (Pipeline): Trained model pipeline.
+        X_test (list of str): Test text data.
+        y_test (list of int): True labels for test data.
+        target_names (list of str): Names of the target categories.
+
+    Returns:
+        None
+    """
+    predicted = model.predict(X_test)
+    print(metrics.classification_report(y_test, predicted, target_names=target_names))
+
+def run_source_classification(categories):
+    """
+    Executes the full classification pipeline:
+    - Loads data for the specified categories
+    - Trains an SVM model
+    - Evaluates model performance
+
+    Args:
+        categories (list of str): List of category names to classify.
+
+    Returns:
+        sklearn.pipeline.Pipeline: The trained model pipeline.
+    """
+    X_train, y_train, X_test, y_test, target_names = load_data(categories)
+    model = train_model(X_train, y_train)
+    evaluate_model(model, X_test, y_test, target_names)
+    return model
+
+if __name__ == "__main__":
+    categories = ['talk.politics.misc', 'rec.sport.hockey', 'sci.space']
+    run_source_classification(categories)
\ No newline at end of file
diff --git a/code/test_source_classifier.py b/code/test_source_classifier.py
new file mode 100644
index 0000000..8ed79a3
--- /dev/null
+++ b/code/test_source_classifier.py
@@ -0,0 +1,71 @@
+# -*- coding: utf-8 -*-
+
+"""
+test_source_classifier.py
+Unit test for svm_source_classifier.py
+Author: Jane Heng
+Date: Oct.10.2025
+"""
+import unittest
+from svm_source_classifier import load_data, train_model, evaluate_model, run_source_classification
+from unittest.mock import patch
+import io
+
+class TestSourceClassifier(unittest.TestCase):
+    """
+    Unit test suite for verifying the functionality of the SVM source classifier pipeline.
+    """
+
+    def setUp(self):
+        """
+        Initializes sample input data and category labels.
+        """
+        self.categories = ['sci.space', 'rec.sport.hockey', 'talk.politics.misc']
+        self.X_sample = [
+            "NASA launched a new satellite.",
+            "The hockey team won the championship.",
+            "Political debates are heating up."
+        ]
+        self.y_sample = [0, 1, 2]
+
+    def test_load_data(self):
+        """
+        Tests whether load_data correctly loads training and testing data.
+        Check if data lengths match and category names.
+        """
+        X_train, y_train, X_test, y_test, target_names = load_data(self.categories)
+        self.assertEqual(len(X_train), len(y_train))
+        self.assertEqual(len(X_test), len(y_test))
+        self.assertEqual(len(target_names), len(self.categories))
+        self.assertIn('sci.space', target_names)
+        print("test_load_data: PASS")
+
+    def test_train_model(self):
+        """
+        Tests if the model is properly trained on sample input.
+        """
+        model = train_model(self.X_sample, self.y_sample)
+        self.assertTrue(hasattr(model, "predict"))
+        print("test_train_model: PASS")
+
+    def test_evaluate_model(self):
+        """
+        Tests whether evaluate_model function.
+        """
+        model = train_model(self.X_sample, self.y_sample)
+        with patch('sys.stdout', new=io.StringIO()) as fake_out:
+            evaluate_model(model, self.X_sample, self.y_sample, self.categories)
+            output = fake_out.getvalue()
+            self.assertIn('sci.space', output)
+        print("test_evaluate_model: PASS")
+
+    def test_run_source_classification(self):
+        """
+        Tests whether run_source_classification function.
+        """
+        model = run_source_classification(self.categories)
+        self.assertTrue(hasattr(model, "predict"))
+        print("test_run_source_classification: PASS")
+
+if __name__ == "__main__":
+    unittest.main()
\ No newline at end of file