Study on Daily Demand Forecasting Orders using Artificial Neural Network
| Autor: | Arthur Arruda Leal Ferreira, Ricardo Pinto Ferreira, Andréa Martiniano, Aleister Ferreira, Renato José Sassi |
|---|---|
| Rok vydání: | 2016 |
| Předmět: |
Engineering
Momentum (technical analysis) General Computer Science Artificial neural network business.industry media_common.quotation_subject 020206 networking & telecommunications Context (language use) Differential (mechanical device) 02 engineering and technology Demand forecasting Machine learning computer.software_genre Network topology Multilayer perceptron Service (economics) 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Artificial intelligence Electrical and Electronic Engineering business computer media_common |
| Zdroj: | IEEE Latin America Transactions. 14:1519-1525 |
| ISSN: | 1548-0992 |
| DOI: | 10.1109/tla.2016.7459644 |
| Popis: | In recent decades, Brazil has undergone several transformations, from a closed economy to a market economy. Transport, processing and distribution of orders remained follow these trends. As a result, the delivery parcel service has become highly complex and competitive. In this context, the forecast demand of orders comes as differential, leading structured productivity and high level of customer service. The paper aims to provide for the daily demand of orders in an Orders Treatment Centre for fifteen days using Artificial Neural Network (ANN). The methodological synthesis of the article is the development of a Artificial Neural Network Multilayer Perceptron (MLP), trained by error back-propagation algorithm. The data for the experiments were collected for 60 days, 45 days to training and 15 days for testing. Experiments were performed with four different topologies of RNA by changing the following parameters: number of hidden layers, number of neurons in the hidden layers, learning rate, momentum rate and stopping criteria. The results obtained with use of RNA in daily demand forecast orders showed good adhesion to the experimental data in the training and testing phases. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|