UTS Praktikum

import pandas as pd
import re
from sklearn.model_selection import train_test_split
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.naive_bayes import MultinomialNB
from sklearn.metrics import classification_report, confusion_matrix, accuracy_score

# 1. baca data

df = pd.read_csv("dataset_uts.csv")
print(df.head())
print(df["label"].value_counts())

# 2. preprocessing sederhana

def preprocess(text):
    text = str(text).lower()
    text = re.sub(r"[^a-zA-Z0-9\s]", " ", text)
    text = re.sub(r"\s+", " ", text).strip()
    return text

df["clean_text"] = df["text"].apply(preprocess)

# 3. split data
X_train, X_test, y_train, y_test = train_test_split(
    df["clean_text"],
    df["label"],
    test_size=0.2,
    random_state=42,
    stratify=df["label"]
)

# 4. ubah teks jadi fitur TF-IDF
vectorizer = TfidfVectorizer(max_features=3000, ngram_range=(1,2))
X_train_tfidf = vectorizer.fit_transform(X_train)
X_test_tfidf = vectorizer.transform(X_test)

# 5. latih model baseline
model = MultinomialNB()
model.fit(X_train_tfidf, y_train)

# 6. evaluasi
pred = model.predict(X_test_tfidf)
print("Accuracy:", accuracy_score(y_test, pred))
print(classification_report(y_test, pred))
print(confusion_matrix(y_test, pred))

from sklearn.linear_model import LogisticRegression
from sklearn.svm import LinearSVC
from sklearn.neighbors import KNeighborsClassifier

models = {
    "Naive Bayes": MultinomialNB(),
    "Logistic Regression": LogisticRegression(max_iter=1000),
    "Linear SVM": LinearSVC(),
    "KNN": KNeighborsClassifier(n_neighbors=3)
}

for nama, clf in models.items():
    clf.fit(X_train_tfidf, y_train)
    pred = clf.predict(X_test_tfidf)
    acc = accuracy_score(y_test, pred)
    print("Model:", nama)
    print("Accuracy:", acc)
    print(classification_report(y_test, pred))
    print("=" * 60)

# contoh struktur simpulan yang bisa ditulis di notebook
"""
Dataset berisi ulasan mahasiswa dengan dua label, yaitu positif dan negatif.
Setelah preprocessing, teks diubah menjadi fitur TF-IDF unigram dan bigram.
Empat model diuji, yaitu Naive Bayes, Logistic Regression, Linear SVM, dan KNN.
Berdasarkan hasil evaluasi, model terbaik adalah .... karena memiliki accuracy dan F1-score tertinggi.
Kesalahan model paling sering muncul pada ulasan yang mengandung pujian dan keluhan sekaligus.
Perbaikan berikutnya adalah menambah data, memperbaiki preprocessing, atau mencoba model transformer.
"""

from transformers import AutoTokenizer, AutoModelForSequenceClassification

model_name = "indobenchmark/indobert-base-p1"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSequenceClassification.from_pretrained(model_name, num_labels=2)

# langkah lanjutan bisa mengikuti pola pada sesi 7:
# 1. ubah DataFrame menjadi Dataset
# 2. tokenisasi
# 3. training dengan Trainer
# 4. evaluasi hasil