Napovedovanje odpovedi izdelkov z metodami strojnega učenja
magistrsko delo Organizacija in management informacijskih sistemov
Abstract
Magistrska naloga obravnava razvoj modela za napovedovanje odpovedi izdelkov v garancijski dobi. Z odpovedovanjem izdelkov in problematiko zagotavljanja popravil v garancijski dobi se soočajo vsa proizvodna podjetja. Zagotavljanje popravil v garancijskem roku podjetjem predstavlja strošek, ki ga poskušajo minimizirati s pomočjo predvidevanja deležev odpovedi. Najpogosteje se napovedi izvedejo z empiričnimi modeli, ki so zgrajeni na preteklih podatkih o podobnih izdelkih in prilagojeni glede na izkušnje. V sklopu magistrske naloge smo s pomočjo različnih metod strojnega učenja in realnih podatkov razvili napovedni model in ocenili uspešnost napovedovanja. Najboljše rezultate napovedovanja smo dobili pri ansamblih regresijskih dreves, pri katerih smo podatke prilagodili eksponentnem modelu. Za zaključek smo pripravili priporočila kateri model uporabiti ob omejenem poznavanju podatkov o odpovedih.
Keywords
napovedni model;odpoved izdelka;garancijski rok;
Data
| Language: | Slovenian |
|---|---|
| Year of publishing: | 2016 |
| Typology: | 2.09 - Master's Thesis |
| Organization: | UM FOV - Faculty of Organizational Sciences |
| Publisher: | [A. Mujanović] |
| UDC: | 004 |
| COBISS: |
7554067
|
| Views: | 1153 |
| Downloads: | 148 |
| Average score: | 0 (0 votes) |
| Metadata: |
|
Other data
| Secondary language: | English |
|---|---|
| Secondary title: | PRODUCT FAILURE PREDICTION WITH MACHINE LEARNING METHODS |
| Secondary abstract: | Master's thesis deals with the development of a model for predicting the failure of products during the warranty period. All of the manufacturing companies are facing the product failure problem and problem with offering the possibility of repairing those products. Providing guarantees represents costs, which companies are trying to minimize by predicting the failure rates. Most often, this is done with empirical models, which are built on historical data for similar products and customized based on experiences. As part of the master's thesis, we developed different models using various methods of machine learning and real data. After development, we assessed the quality of prediction for each model. Ensembles of regression trees obtained the best results; in that case, the data was fitted to the exponential model. To wind up, we prepared recommendations which model to use in different scenarios. |
| Secondary keywords: | Prediction model;product failure;warranty period;quality;machine learning;neural networks;regression trees.; |
| URN: | URN:SI:UM: |
| Type (COBISS): | Master's thesis/paper |
| Thesis comment: | Univ. v Mariboru, Fak. za organizacijske vede |
| Pages: | 52 f. |
| ID: | 9128242 |
Recommended works:
Napovedovanje odpovedi izdelkov z metodami strojnega učenja
2016,
magistrsko delo Organizacija in management informacijskih sistemov
Napovedovanje odpovedi izdelkov z metodami globokega učenja
2017,
diplomsko delo univerzitetnega študija Organizacija in management informacijskih sistemov
Napovedovanje odpovedi stroja za proizvodnjo tesnil z metodami strojnega učenja
2018,
no subtitle data available
Štetje množic z metodami strojnega učenja
2021,
diplomsko delo
Churn prediction with machine learning methods
2019,
no subtitle data available