메뉴
Use when writing or reviewing pandas code. Covers idiomatic, vectorized, memory-efficient patterns; avoiding SettingWithCopyWarning, chained indexing, and slow apply loops.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Use when preparing raw data for modeling — handling missing values, duplicates, inconsistent types, outliers, and bad categorical values. Emphasizes fitting all imputation on train-only to avoid leakage.
Use when running ML experiments that need to be compared, reproduced, or shared. Covers MLflow/Weights & Biases logging, what to track, run organization, and model registry basics.
Use when starting on a new dataset, before modeling, or when asked to "explore", "profile", "understand", or "summarize" data. Covers structured EDA, distributions, correlations, target leakage checks, and visualization.
Use when creating, encoding, scaling, or selecting features for ML models. Covers categorical encoding, numeric transforms, datetime/text/aggregation features, and leakage-safe target encoding.
Use when optimizing model hyperparameters. Covers search strategy (random vs Bayesian/Optuna), leakage-safe tuning inside CV, search-space design, early stopping, and budget management.
Use when the target is rare (fraud, churn, disease, anomalies). Covers correct metrics, resampling (SMOTE/undersampling), class weights, threshold tuning, and avoiding the accuracy trap and resampling leakage.
| name | pandas-patterns |
| description | Use when writing or reviewing pandas code. Covers idiomatic, vectorized, memory-efficient patterns; avoiding SettingWithCopyWarning, chained indexing, and slow apply loops. |
Most pandas pain comes from three things: chained indexing, row-wise apply, and ignoring dtypes/memory. This skill encodes the idioms that keep pandas correct and fast.
SettingWithCopyWarning..loc, never chained.
df.loc[df["age"] > 30, "segment"] = "senior" # correct
# df[df["age"] > 30]["segment"] = "senior" # WRONG: SettingWithCopyWarning, no-op risk
apply(axis=1). Row-wise apply is a Python loop.
df["bmi"] = df["weight"] / df["height"] ** 2 # fast
# df.apply(lambda r: r.weight / r.height**2, axis=1) # 100x slower
np.select / np.where for conditional columns.
import numpy as np
df["tier"] = np.select(
[df.spend > 1000, df.spend > 100],
["gold", "silver"],
default="bronze",
)
category for low-cardinality strings, int32/float32 where safe.
df["country"] = df["country"].astype("category")
merge over loops for joins, and validate join cardinality:
df = orders.merge(users, on="user_id", how="left", validate="m:1")
df.groupby(..., observed=True).agg(...) — observed=True avoids exploding categorical combinations.pd.eval / df.query() for large boolean filters.chunksize=) or switch to Polars/DuckDB when pandas is the bottleneck.df.pipe(fn) to compose transformations without intermediate variables.result = (
df
.query("status == 'active'")
.assign(revenue=lambda d: d.qty * d.price)
.groupby("region", observed=True)
.agg(total=("revenue", "sum"))
.reset_index()
)
inplace=True rarely saves memory and breaks chaining — avoid it.iterrows — almost always replaceable with vectorization or itertuples.Int64 if you must keep integers.Clean, vectorized transformations that downstream skills (feature-engineering, model-evaluation) can run quickly on full datasets.