FORECASTING & PLANNING · SC-12
Multi-Horizon Demand Forecasting & Inventory Planning
A forecasting system that compares baselines and machine-learning models across multiple planning horizons.
Starting point
Aggregate accuracy can hide weak planning performance and lead to the wrong stock, capacity or financial action.
01 · BUSINESS PROBLEM
A forecast only matters at the horizon where a decision is made.
Aggregate accuracy can hide weak planning performance and lead to the wrong stock, capacity or financial action.
Hidden bias
Average error can look acceptable while direction is systematically wrong.
Wrong horizon
A model can be useful short term and weak for longer planning.
No decision link
Accuracy without an operating action creates little value.
02 · DECISION LOGIC
Evaluate the model where the business uses it.
Forecast quality is separated by horizon and compared with simple baselines.
Decision sequence
Each stage answers a different operating question
Immediate operating signal.
Planning before uncertainty compounds.
Model degradation becomes visible.
Usefulness is judged as its own decision.
Decision rule — Model complexity only matters if it changes a planning decision.
03 · WHAT CHANGED
From forecast quality to planning action.
The system connects prediction, backtesting, scenario assumptions and the decision layer.
Benchmark against simple statistical baselines.
Evaluate error and bias by horizon.
Expose scenarios and planning assumptions.
Translate outputs into stock, capacity or financial decisions.
04 · ARCHITECTURE
A modular path from input to decision.
Inputs → preparation → core logic → validation → decision output
SC-12 · SYSTEM ARCHITECTURE
Inputs → preparation → core logic → validation → decision output
Public portfolio implementation
Inputs
Source signals
Capture the operating inputs required by the system. [Python]
Preparation layer
Normalize context and create a stable analytical contract. [Statsmodels]
Core system
Core engine
Run the main analytical or automation logic. [Machine Learning]
Decision logic
Apply the rule, model or orchestration logic that changes the decision. [FastAPI]
Validation
Validation
Test outputs against explicit quality criteria. [PostgreSQL]
Controls
Keep approvals, thresholds or constraints visible. [Prefect]
Decision output
Decision output
Expose the result in a form the user can act on.
Monitoring
Record outcomes, exceptions and evidence for iteration.
Integration boundaries
Python
Defined responsibility inside the system; replaceable if another tool fits the requirement better.
Statsmodels
Defined responsibility inside the system; replaceable if another tool fits the requirement better.
Machine Learning
Defined responsibility inside the system; replaceable if another tool fits the requirement better.
05 · EVIDENCE & ECONOMICS
Measure what changes the decision.
Public implementation, inspectable technical proof and decision-focused validation.
Model evaluation
WAPE
6.45%
Representative public example.
Bias
-6.45%
Representative public example.
Horizons
4
Representative public example.
Planning coverage
Scenarios
3
Representative public example.
Validation checks
12
Representative public example.
Reference economics
€2.0M
Inventory base
5%
Excess reduction
€100k
Capital released
06 · TECHNICAL PROOF
Review the code behind the project.
Tools used
BUSINESS CONCLUSION
Forecasting creates value when it changes the plan.
The useful output is not the prediction itself; it is a better decision made with explicit uncertainty.