菜单
Generate lag features for multiple targets over N days by shifting evaluation dates and self-joining per entity, creating a wide feature matrix of past target values
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Memory-efficient Keras generator that streams sharded CSV files in chunks, renders strokes to images on-the-fly, and yields batches for training on datasets too large for memory
Stack 1D convolutions for local feature extraction before bidirectional LSTMs to classify variable-length stroke sequences into hundreds of doodle categories
Partition a large dataset into N balanced shards using integer key modulo arithmetic for reproducible, class-interleaved splits across CSV files
Normalize raw stroke coordinates to 0-255 range, resample at uniform arc-length spacing, then apply Ramer-Douglas-Peucker simplification
Render stroke sequences to grayscale images with temporal intensity encoding where earlier strokes are brighter and later strokes fade to encode drawing order
Score each face in a video as anomalous by computing L2 distance from the embedding centroid of all faces, then convert to probability via logistic function
| name | timeseries-multi-lag-target-features |
| description | Generate lag features for multiple targets over N days by shifting evaluation dates and self-joining per entity, creating a wide feature matrix of past target values |
For entity-level time-series forecasting with multiple targets, lagging each target by 1..N days and pivoting into separate columns creates a dense feature matrix capturing recent history. The self-join approach shifts the date column by the lag amount and merges back on (entity_id, date), producing columns like target1_1, target1_2, ..., target4_20 — 80 features for 4 targets x 20 lags.
from datetime import timedelta
TARGETS = ["target1", "target2", "target3", "target4"]
LAGS = list(range(1, 21))
def add_lag(df, lag):
lagged = df[["entity_id", "date"] + TARGETS].copy()
lagged["date"] = lagged["date"] + timedelta(days=lag)
df = df.merge(lagged, on=["entity_id", "date"], suffixes=("", f"_{lag}"), how="left")
return df
for lag in LAGS:
df = add_lag(df, lag)
lag_cols = [f"{t}_{l}" for l in reversed(LAGS) for t in TARGETS]
shift() within groups instead of merge