RiskFlow Engine v1.0
数桥风险引擎 v1.0
Data Bridge ConsultingSupply Chain — Predicting Supplier Disruption Risk
This demo estimates the forward-looking risk that a supplier will experience a disruption within a chosen prediction horizon. A disruption is defined as default, production halt, or a delivery delay exceeding 30 days. It illustrates how procurement and finance teams can use early warning signals to strengthen supply chain resilience and working-capital management.
What this demo does
- Goal: identify unstable suppliers early so teams can adjust orders, diversify sources, or build buffer stock.
- Data: monthly synthetic supplier records combining operational metrics, financial stress indicators and macro factors.
- Approach: RiskFlow learns how patterns in deliveries, complaints, leverage and macro pressure relate to future disruption risk at different prediction horizons.
- Alert rule (demo): raise a warning when predicted disruption risk stays above a chosen threshold for at least two consecutive months.
- Lead time: the number of months between the first alert and the month when disruption occurs.
Dataset overview
The data in this demo are fully synthetic and designed to resemble a monthly supplier panel used in real-world supply-chain monitoring. Each supplier has a history of delivery performance, financial ratios and region-level risk indicators, combined with time-varying macro and logistics indices such as trade tension and shipping costs. The dataset is split into a training sample and a held-out test sample so that model behaviour can be illustrated without exposing any real supplier information.
Variables
| Name | Type | Description |
|---|---|---|
Delivery rate
|
continuous | On-time delivery ratio from 0–1; tends to decline before a disruption. |
Inventory turnover
|
continuous | How quickly inventory is sold or used; low values may signal weak demand or excess stock. |
Leverage ratio
|
continuous | Debt-to-assets ratio; higher leverage indicates greater financial fragility. |
Liquidity ratio
|
continuous | Current assets divided by current liabilities; lower values point to short-term cash stress. |
Order backlog
|
continuous | Normalised backlog pressure; high levels can indicate capacity bottlenecks. |
FX exposure
|
continuous | Share of revenue in foreign currency; exposes the supplier to exchange-rate swings. |
Complaint index
|
continuous | Normalised complaint rate from 0–1; a rising trend is often an early warning sign. |
Region risk score
|
continuous | Baseline macro and geopolitical risk level for the supplier’s region. |
Global trade tension
|
continuous | Synthetic index of trade and policy frictions; higher values mean a tougher trade environment. |
Shipping cost index
|
continuous | Freight and shipping cost pressure; higher values reflect more expensive logistics. |
Commodity price index
|
continuous | Upstream commodity cost pressure; higher values can squeeze supplier margins. |
FX volatility
|
continuous | Broad exchange-rate volatility index; higher levels increase uncertainty for exposed suppliers. |
Region
|
categorical | Supplier’s region category (e.g. CN, EU, US, ASEAN, LATAM, AFR, IND). |
target
|
binary | 0 = Stable, 1 = Disrupted |
Key definitions
| Term | Description |
|---|---|
Disruption event
|
Default, production halt, or delivery delay > 30 days (target=1 at event time period). |
id, time
|
Supplier id and sequential time index within the observed window. |
start_time, end_time
|
First and last observed time per supplier (rows outside are zeroed). |
Target distribution
Download demo data
Download ZIPWhat you’ll see
- How financial and operational drivers change in importance across different prediction horizons.
- ROC curves across horizons and interactive Top-10 high-risk supplier lists by time period and prediction horizon.