Feature platforms and point-in-time correctness
Use feature platforms and point-in-time correctness to move the ml platforms production brief toward a defensible release.
A late feature silently corrupts every downstream prediction.
This lesson isolates feature platforms and point-in-time correctness as one decision inside that system. The people affected are product users and operators; the learning data must carry event time, availability time, ownership, and version; and the operating envelope is a declared latency, cost, and operator-capacity budget.
Choose whether and how to use watermarks at a declared prediction cutoff.
Measure decision utility alongside calibration, slice reliability, and system latency—not model score alone.
A late feature silently corrupts every downstream prediction. An unsafe release must degrade to a named baseline or the last known-good version.
Build the mental model before the machinery.
The core move is to treat feature platforms and point-in-time correctness as a contract between data, a computation, and an action. Make temporal correctness, ownership, lineage, reproducibility, and escape hatches the platform defaults. The implementation becomes easier to debug once you can state which inputs exist, which state is learned, what output means, and what must remain invariant after serialization.
watermarks
Define it in a hand-checkable form and name the prediction-time inputs.
feature stores
Connect it to the production metric and identify what it cannot guarantee.
Airflow/Dagster
Stress it with a slice, a temporal boundary, and a failure-safe alternative.
MLflow
Stress it with a slice, a temporal boundary, and a failure-safe alternative.
Lessons 1, 2 in this course.
Name every symbol. Check every shape.
Point-in-time feature lookup is the central invariant for this lesson. The formula is useful only when its inputs match the production cutoff and its output maps to an action.
Point-in-time feature lookup
| Symbol | Meaning / shape / unit |
|---|---|
j | entity |
t_p | historical prediction time |
e | eligible event |
Open derivation and numerical substitution
Start from the production quantity being optimized, substitute the observed values with their declared units, then isolate the model-controlled term. Preserve shape annotations at each step so broadcasting or aggregation cannot silently change the result.
- Write the named inputs: j, t_p, e.
- Substitute one small, hand-checkable batch before vectorizing.
- Calculate an independent reference value and compare within a declared tolerance.
# equation → code contract
inputs = validate_shapes_and_units(batch)
value = compute_c17(inputs)
assert is_finite(value)Calculate it small. Shape it realistically. Break it on purpose.
A result you can reproduce on paper
A Monday refund arriving Wednesday must not appear in a Tuesday training row.
- Write every input and unit.
- Substitute values into the point-in-time feature lookup equation above.
- Compare the result to one simple baseline and explain the direction of the difference.
The same reasoning under real constraints
Kafka feeds Flink aggregates online while immutable columnar data supports matching historical joins.
The production record includes the data snapshot, transformation state, artifact identity, cutoff, score, decision, and the version of the policy that consumed it.
The attractive result you should reject
A backfill uses corrected future knowledge and silently inflates offline AUC.
Diagnostic: replay the smallest failing slice from immutable inputs, then compare each boundary rather than retuning the model.
Change one assumption and make the tradeoff visible.
This lab runs predefined TypeScript only. It never executes learner code. Use the slider, numeric input, reset, live text, or table—the computation is the same.
Point-in-time freshness lab
Delay feature arrival and see which historical rows remain eligible.
Assumption: Five events have fixed prediction cutoffs separated by six hours.
Open nonvisual data table
| Item | Computed state | Interpretation |
|---|---|---|
| row 1 | cutoff +0h | not yet knowable |
| row 2 | cutoff +6h | eligible |
| row 3 | cutoff +12h | eligible |
| row 4 | cutoff +18h | eligible |
| row 5 | cutoff +24h | eligible |
Trace the complete operating path.
- 01
Validate and version watermarks.
- 02
Compute feature platforms and point-in-time correctness from prediction-time-safe inputs.
- 03
Persist model, feature, and configuration identities together.
- 04
Serve or materialize behind explicit a declared latency, cost, and operator-capacity budget.
- 05
Join telemetry to mature outcomes and retain a rollback path.
Observability
Join service health, input quality, prediction distributions, slice behavior, and mature outcomes by exact version.
Cost
Measure storage, preprocessing, compute, queueing, and human review under a representative arrival pattern.
Failure modes
A backfill uses corrected future knowledge and silently inflates offline AUC. Add a detector, owner, mitigation, and stop condition for this class of failure.
Alternatives
Compare a rule, a simpler statistical baseline, and a different system boundary before adding model complexity.
Explain the contract, not just the vocabulary.
Local mini ML platform
Specify idempotent events, a point-in-time join, shared feature definition, quality gates, lineage, and model registration.
- Data contract
- Late-event semantics
- Feature parity design
- Lineage/registry record
Read primary material with a purpose.
Return to the opening failure.
Make temporal correctness, ownership, lineage, reproducibility, and escape hatches the platform defaults.
For this lesson, the release evidence is a hand-checked formal result, deterministic simulation output, a ≥80% checkpoint, the production rubric, and a named fallback. The course resolves when the system can produce a traced batch/stream platform with orchestration, registry, lineage, and ownership.