NLP Academy · 课时

端到端使用 scikit-learn 流水线

整洁地串联向量化器与模型

端到端使用 scikit-learn 流水线 是 CoddyKit 上的免费 NLP Academy 课时。这是第 2 节课，共 4 节。你可以在下方免费阅读本课时的完整内容 — 然后在浏览器中使用内置代码编辑器和全天候 AI 导师进行实践。这是 NLP Academy 学习路径的一部分，你的进度在网页和 CoddyKit 应用中同步。 NLP Academy 课程共包含 4 节课。

本课时的部分内容尚未翻译，以英文显示。

What Is a Pipeline?

A pipeline chains your steps into one object: clean, vectorize, then classify. Call it once and every step runs in the right order. 🔗

Why Glue Steps Together

Doing vectorizing and training by hand invites bugs. A pipeline keeps the steps locked together so they never drift out of sync.

Import the Pieces

You need a vectorizer and a classifier. Import the Pipeline class plus the two components you want to chain.

from sklearn.pipeline import Pipeline
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.linear_model import LogisticRegression

Build the Pipeline

List your steps as named pairs. The first transforms text into features, the last makes the prediction.

pipe = Pipeline([
    ("tfidf", TfidfVectorizer()),
    ("clf", LogisticRegression())
])

Fit on Raw Text

Call fit with your raw text and labels. The pipeline vectorizes and trains in a single line, no manual steps.

pipe.fit(X_train, y_train)

Predict in One Call

To classify new text, call predict. The same vectorizer is reused automatically, so train and test stay perfectly consistent.

preds = pipe.predict(X_test)

No Data Leakage

Pipelines learn the vocabulary only from training data. This stops data leakage, where test info sneaks into training and inflates your score.

Score the Pipeline

Use score to get accuracy on held-out data in one call. The pipeline handles vectorizing the test text for you.

acc = pipe.score(X_test, y_test)
print(acc)

Tune the Whole Chain

Grid search can tune any step. Name parameters with the step name plus double underscore to reach inside a component.

params = {"tfidf__ngram_range": [(1,1), (1,2)],
          "clf__C": [0.1, 1, 10]}

Cross-Validate Safely

Pass the whole pipeline into cross-validation. Each fold re-fits the vectorizer, so every score reflects truly unseen data.

from sklearn.model_selection import cross_val_score
scores = cross_val_score(pipe, X, y, cv=5)

One Object to Ship

The best part: a fitted pipeline is a single object. Save and load it as one unit, and inference matches training exactly.

Quick Check

Think about the main safety benefit a pipeline gives you.

Recap

You chained a vectorizer and classifier into one pipeline, fit on raw text, predicted in a line, tuned the chain, and avoided leakage. 🎉

常见问题解答

「端到端使用 scikit-learn 流水线」课时是免费的吗？

是的 — 「端到端使用 scikit-learn 流水线」的完整文本可在网页上免费阅读。要进行交互式练习（内置代码编辑器和全天候 AI 导师）并解锁 NLP Academy 课程的其余内容，请升级到 CoddyKit PRO。 NLP Academy 课程共包含 4 节课。

「端到端使用 scikit-learn 流水线」这节课中我会学到什么？

整洁地串联向量化器与模型你通过在浏览器中直接运行的动手代码来练习 NLP Academy，全天候 AI 导师会在你学习这节课的过程中回答你的问题。

学习 NLP Academy 需要有经验吗？

无需任何先前经验。CoddyKit 上的 NLP Academy 课程适合初学者到高级学习者，你可以从这里开始或从头开始，按照自己的节奏学习。这是第 2 节课，共 4 节。

「端到端使用 scikit-learn 流水线」课时需要多长时间？

大多数 CoddyKit 课程大约需要 5–10 分钟。每节课都很精短且互动，所以你能稳步进步，并在网页和应用中从离开的地方继续。

我能在这节 NLP Academy 课中编写并运行代码吗？

能。每节 NLP Academy 课都包含内置代码编辑器，你可以在浏览器中直接编写并运行真实代码，并获得即时 AI 反馈 — 无需本地设置。

此课程中的所有课时

← 返回 NLP Academy