Data Science Academy · 课时

使用 transform 创建分组特征

将分组统计值对齐回各行

使用 transform 创建分组特征 是 CoddyKit 上的免费 Data Science Academy 课时。这是第 4 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 Data Science Academy 学习路径的一部分，你的进度在网页和 CoddyKit 应用中同步。 Data Science 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。

agg Shrinks, transform Keeps

Where agg returns one row per group, transform returns a value for every original row, same length as the input.

Why That Matters

Because transform keeps the original shape, you can drop its result straight back into your DataFrame as a new column.

df["city_avg"] = df.groupby("city")["sales"].transform("mean")

Broadcast a Group Stat

Each row receives its own group statistic, so every Paris row gets the Paris average, neatly aligned.

Build a Group-Relative Feature

Subtract the group mean to see how far each row sits from its group average, a powerful model feature.

df["diff"] = df["sales"] - df.groupby("city")["sales"].transform("mean")

Normalize Within Groups

Divide by a group total to turn raw counts into a within-group share that compares fairly across groups.

df["share"] = df["sales"] / df.groupby("city")["sales"].transform("sum")

Fill Missing With Group Means

A clean imputation trick: replace gaps using each row group mean instead of one global number. 🧩

df["sales"] = df.groupby("city")["sales"].transform(lambda s: s.fillna(s.mean()))

Use Strings or Functions

Like agg, transform accepts a function name such as "mean" or your own callable returning a like-shaped result.

g.transform("sum")

It Preserves the Row Index

The output keeps the original index, which is why assigning it back as a column lines up perfectly.

Rank Within Each Group

Combine transform with rank to score each row against its group, great for top-N-per-group logic.

df["rk"] = df.groupby("city")["sales"].rank(ascending=False)

transform vs apply

Choose transform when output must match input length; reach for apply only when you need full flexibility.

A Common Pitfall

Your function must return a result the same length as the group, or transform raises an error instead of guessing.

Quick Check

What makes transform different from agg?

Recap: Features That Fit Right Back

You can now compute group-wise features that keep the original shape, perfect for normalizing, imputing, and ranking. 🚀

常见问题解答

「使用 transform 创建分组特征」课时是免费的吗？

是的 — 「使用 transform 创建分组特征」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 Data Science Academy 课程的其余内容，请升级到 CoddyKit PRO。 Data Science Academy 课程共包含 4 节课。

「使用 transform 创建分组特征」这节课中我会学到什么？

将分组统计值对齐回各行你通过在浏览器中直接运行的动手代码来练习 Data Science 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 Data Science Academy 需要有经验吗？

无需任何先前经验。CoddyKit 上的 Data Science 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 4 节课，共 4 节。

「使用 transform 创建分组特征」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 Data Science Academy 课中编写并运行代码吗？

能。每节 Data Science 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 Data Science Academy