SMART PROCUREMENT / WORKING PROTOTYPES

Change the brief.
Understand the prediction.

Four browser tools inspired by my capstone: PO delays, approval bottlenecks, unusual purchases and carbon impact. Enter a fictional scenario, change the assumptions and inspect why the result changes.

New synthetic demonstrators. These are newly fitted public models, not the original dissertation models or an operational university service. Inputs remain in your browser; no supplier, employee or institutional records are requested.

Loading four model prototypes…

PO Delay Predictor

Explore a fictional requisition’s delay score.

Approval Delay Predictor

Compare possible workflow bottlenecks.

Anomaly / Fraud Review

Screen unusual patterns for human review, not accusations.

Carbon Impact Predictor

Explore a hypothetical category and freight footprint.

Explore the public capstone summary · Explore threshold costs · Check data readiness

Personal tool suggestions

Enable JavaScript to choose optional feedback-based suggestions.

EXPLAINABLE AI & DATA / READ THE METHOD

From a requisition to an explained estimate.

These browser experiments use exported models trained on synthetic procurement scenarios. They demonstrate model behaviour, not live organisational forecasts.

Follow the reasoning.

GenerateSynthetic requisition scenarios
FitPython training and held-out checks
PredictExported model calculations in the browser
QuestionCompare errors, thresholds and operational consequences

Explore the building blocks.

AI Capability4

AI/Data Concepts1

Algorithm4

Evaluation & XAI4

Python Package2

Responsible AI1

Number of concepts in this guide, not performance or usage measures.

Explore the explanation matrix · 16 concepts

Concept categoryFind a term or example

16 concepts shown

On small screens, scroll the table sideways. “Learning context” explains a related idea; it does not mean that method runs on this page. Deep learning is the standard term for “deep ML”.

Technical term	Technical description / definition	In plain English	How this page uses it
Synthetic dataAI/Data Concepts	Generated records used to represent scenarios rather than directly report observed individuals or events.Primary reference ↗	A practice dataset. Useful for demonstration, not proof of workplace impact.	Demonstration The lab uses generated procurement records; no employer purchase orders are queried.
Random forestAlgorithm	An ensemble of decision trees whose outputs are combined for classification or regression.Primary reference ↗	Several branching rulebooks contribute to one estimate.	Running here The delay classifier combines tree probabilities from its exported forest.
Gradient boostingAlgorithm	An additive ensemble trained sequentially to improve a specified loss objective.Primary reference ↗	Build a sequence of small models that work on what earlier ones missed.	Running here The approval classifier combines the exported boosting trees and class scores.
Isolation forestAlgorithm	An anomaly detector based on how quickly random partition trees isolate an observation.Primary reference ↗	Flag unusual records for review; unusual does not mean fraudulent or wrong.	Running here Unusual procurement cases are scored by tree path lengths for review, not labelled wrongdoing.
Ridge regressionAlgorithm	A linear regression model with an L2 coefficient penalty to limit excessive parameter magnitude.Primary reference ↗	Fit a trend while discouraging overly extreme weights.	Running here The linear model estimates the lab’s continuous outcome from its stated inputs.
Decision thresholdEvaluation & XAI	A cutoff that converts a model score into an action or predicted category.Primary reference ↗	Decide how strong a signal must be before you act; changing it changes mistakes.	Running here Changing the delay threshold changes which synthetic cases trigger an alert.
Confusion matrix, precision and recallEvaluation & XAI	A confusion matrix counts predicted versus actual classes. Precision measures positive-result reliability; recall measures positive-case coverage.Primary reference ↗	Which alarms were right, which were false, and which cases were missed?	Demonstration Evaluation charts expose false positives and missed delays on the lab’s test data.
Residuals and prediction errorEvaluation & XAI	A residual compares an observed value with its fitted or predicted value. Error metrics summarise those discrepancies.Primary reference ↗	How far did the estimate miss, and where does it miss most?	Demonstration Continuous-outcome errors show the gap between synthetic references and predictions.
Held-out evaluationEvaluation & XAI	Assessment on examples excluded from fitting and model selection, according to the declared split.Primary reference ↗	Keep some answers out of practice so the test means something.	Published analysis The exported training workflow separates model fitting and test examples.
NumPyPython Package	A Python library for multidimensional arrays and numerical operations.Primary reference ↗	The numerical toolbox behind many Python calculations.	Build-time Python NumPy supports synthetic dataset creation and numerical calculations.
scikit-learnPython Package	A Python library providing estimators, preprocessing tools and model-evaluation utilities.Primary reference ↗	A toolkit of models and fairer ways to test them.	Build-time Python The training source uses RandomForestClassifier, GradientBoostingClassifier, IsolationForest and Ridge.
Human oversight and consentResponsible AI	People retain appropriate review authority, permission controls and correction routes for data use and outputs.Primary reference ↗	Keep someone able to question, correct or stop the result.	Responsible use An estimate supports investigation; it does not approve spending or replace procurement judgement.
Machine learning (ML)AI Capability	Methods that estimate patterns or parameters from examples to support inference on other inputs.Primary reference ↗	Learn a pattern from examples, then check whether it helps with new ones.	Learning context ML learns from examples. Read the page-specific methods below to see whether any model runs here.
Deep learningAI Capability	Machine learning using neural networks with multiple representation-learning layers.Primary reference ↗	A layered model learns useful representations; more layers do not guarantee a better answer.	Learning context Deep learning is an AI capability, not a label for every chart. No deep model is implied by this explanation panel.
Natural language processing (NLP)AI Capability	Computational methods for representing, analysing or generating human language.Primary reference ↗	Help a computer work with words, while remembering that context matters.	Learning context Text search, translation, thematic classification and generation are different language tasks; their implementation is identified separately.
Sentiment analysisAI Capability	A language-analysis task that estimates an expressed attitude, often as polarity or sentiment categories.Primary reference ↗	Estimate the tone of words—not the private feelings or personality of their author.	Learning context Sentiment estimates expressed tone. It must not be confused with topic classification or a diagnosis of a person.

Change the brief.Understand the prediction.