Module auto_features 

Automatic feature generation for time-series data.

Produces a FeatureMatrix containing named feature columns derived from raw timestamps and values:

• Temporal: hour_of_day, day_of_week, day_of_month, month_of_year, is_weekend
• Statistical: rolling_mean, rolling_std, rolling_min, rolling_max at configurable windows (default: 60, 360, 1440, 10080 — representing 1 h, 6 h, 24 h, 7 d when the interval is 1 minute).
• Lag-based: lag_1, lag_2, lag_period (requires known period)

Optional variance-threshold pruning removes near-constant features.

Structs

AutoFeatureConfig

Configuration for auto-feature generation.

FeatureMatrix

Named feature matrix: one row per observation, one column per feature.

Functions

auto_features

Generate features from timestamps (nanosecond epoch) and values.