基于Scikit-learn SVM训练分类器：未见过数据分类求助

Alright, here's a clean, structured breakdown of your text classification workflow formatted properly with Markdown:

1. Sample Dataset #

Below is your labeled training data presented in a readable table:

2. Model Selection & Implementation #

You tested two standard multi-class text classifiers and settled on Support Vector Machines (SVM) over MultinomialNB for better fit. Here’s a practical code snippet to implement and fit the SVM model using scikit-learn:

from sklearn.svm import SVC
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.pipeline import Pipeline
from sklearn.model_selection import train_test_split

# Prepare your dataset (adjust based on how you're storing your data)
X = ["inadequate stock", "Too many failures", "no documentation", "good product", "long delivery times"]
y = ["Availability", "Quality", "Customer Service", "Satisfied", "Delivery"]

# Split into training/validation sets (good practice for evaluation)
X_train, X_val, y_train, y_val = train_test_split(X, y, test_size=0.2, random_state=42)

# Build a end-to-end pipeline: TF-IDF text vectorization + SVM classifier
text_classifier = Pipeline([
    ('tfidf', TfidfVectorizer(stop_words='english')),  # Remove common stopwords for cleaner features
    ('svm', SVC(kernel='linear', C=1.0))  # Linear kernel is generally effective for text tasks
])

# Train the model on your labeled data
text_classifier.fit(X_train, y_train)

# Classify unseen data with the trained model
unseen_comments = ["my order took 2 weeks to arrive", "item stopped working after 3 days"]
predicted_categories = text_classifier.predict(unseen_comments)
print(predicted_categories)

Pro tip: To optimize performance, consider tuning hyperparameters like C (regularization strength for SVM) or n-gram ranges in TF-IDF using cross-validation.

内容的提问来源于stack exchange，提问作者Chat Peters