Estimation de la distribution de la demande pour des familles de produits textiles
Bibliographic record
Abstract
RÉSUMÉ: L’évolution rapide de notre population, de nos technologies, de notre climat et de nos besoins nous incite à développer des stratégies d’adaptation à un rythme toujours plus soutenu. La prédiction est un outil qui a le potentiel de faciliter notre progression dans l’environnement incertain dans lequel nous vivons. C’est la raison pour laquelle de nombreuses entreprises comme Logistik Unicorp cherchent à développer des modèles qui leur permettent de prédire la demande de leurs clients afin d’ajuster leur production à cette dernière. Logistik Unicorp est une entreprise québécoise qui se spécialise dans la production d’uniformes, qu’elle distribue aux quatre coins du globe. Désireuse d’améliorer sa gestion d’inventaire, l’organisation souhaite développer un modèle qui lui permette d’obtenir des distributions prédictives de la demande hebdomadaire et mensuelle de ses clients. Ce mémoire présente une méthode qui exploite des modèles bayésiens descriptifs pour obtenir les prédictions d’une famille de produits dont la demande présente une importante saisonnalité. La simulation par chaînes de Markov est utilisée pour prédire l’incertitude des données résiduelles de la demande, tandis que les composantes saisonnière et tendancielle sont estimées à partir des données historiques. Les prédictions produites sont comparées à celles que peuvent nous offrir des outils comme SARIMA et les modèles d’espaces-états. Les résultats démontrent que la méthode produit des résultats semblables à ceux que nous offrent nos deux outils de référence. Cependant, la méthode proposée, qui fournit distribution prédictive approximative de la demande, présente un potentiel d’amélioration intéressant vis-à-vis de la gestion de l’approvisionnement du partenaire. Cela car elle lui permettra d’ajuster sa demande avec une estimation probabiliste des risques de rupture de stock. ABSTRACT: The rapid evolution of our population, our technologies, our climate and our needs is prompting us to develop adaptation strategies at an ever-increasing pace. Prediction is a tool that has the potential to facilitate our progress in the uncertain environment in which we live. That's why many companies like Logistik Unicorp are seeking to develop models that enable them to predict customer demand and adjust their production accordingly. Logistik Unicorp is a Quebec-based company specialising in the production of uniforms, which it distributes to the four corners of the globe. Wishing to improve its inventory management, the organisation wanted to develop a model that would enable it to obtain predictive distributions of its customers' weekly and monthly demand. This dissertation presents a method that exploits descriptive Bayesian models to obtain predictions for a family of products whose demand exhibits significant seasonality. Markov chain simulation is used to predict the uncertainty in the residual demand data, while the seasonal and trend components are estimated from historical data. The predictions produced are compared with those offered by tools such as SARIMA and state-space models. The results show that the method produces results similar to those offered by our two reference tools. However, the proposed method, which provides an approximate predictive distribution of demand, has an interesting potential for improvement with regard to the partner's supply management. This is because it will enable them to adjust their demand with a probabilistic estimate of the risk of stock shortages.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.004 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.004 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".