RiskFlow Engine v1.0
数桥风险引擎 v1.0
Data Bridge ConsultingBring Your Own Data (BYOD)
BYOD lets you quickly test how RiskFlow behaves on your own dataset, before committing to a full Proof of Concept (PoC) or deployment.
How it works (workflow + data requirements)
- Step 1 — Define the risk event and choose a forecasting horizon.
- Step 2 — Prepare a small, verifiable dataset.
- Step 3 — Ensure you have entity IDs, time column, and feature columns.
- Step 4 — Run the preview here, and review outputs + diagnostics.
- Data privacy — Uploaded datasets are processed locally for this session only and are not viewed, stored, or reused by us.
Recommended constraints: CSV format • up to 50,000 rows • up to 50 MB • less than 12 features. For this preview upload, we recommend ≤ 5,000 rows to keep it fast.
Expected data format (CSV, long/panel)
Download templatesUpload a dataset in long (panel) format, where each row represents one entity at one point in time. Column names are required. Order is recommended.
| Column | Required | Description | Example |
|---|---|---|---|
entity_id |
Yes | Unique identifier for each entity (firm / supplier / patient, etc.). | A001 |
time |
Yes |
Time index for each row. Supported formats: YYYY-MM, YYYY-MM-DD, or a numeric index 1,2,3….
|
2025-02 |
target |
Yes | Event label (0/1). | 0 / 1 |
feat* |
Yes |
Feature columns (numeric recommended). Example: feat1, leverage, liquidity…
|
0.12 |
Use case examples (targets & feature ideas)
Select a common BYOD scenario to view suggested target definitions, feature ideas, and notes.
Upload dataset & run (Preview)
Upload a small, anonymised sample to validate feasibility. This preview runs a lightweight check and prepares the data for modelling.
Need help with your dataset?
If you are unsure about the event definition, the time field, or the format mapping, email us and we’ll help you validate the dataset quickly.