Triastuti Wuryandari
Departemen Statistika, Fakultas Sains Dan Matematika, Universitas Diponegoro

Published : 28 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search
Journal : MATEMATIKA

PEMILIHAN ARSITEKTUR OPTIMAL MODEL NN DENGAN METODE KONTRIBUSI INCREMENT Wuryandari, Triastuti
MATEMATIKA Vol 9, No 3 (2006): JURNAL MATEMATIKA
Publisher : MATEMATIKA

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (175.6 KB)

Abstract

Neural Network is an information processing system that has certain characteristic in common with biological neural network. In development NN has been many applied in several surface, one of them is for forecasting. For the best application of NN, architecture has determined. One of methode to get optimal architecture NN is incremental contribution methods. This methods will to determine the size of hidden and input cell in the network with excluding respectively. One of the unit cell with a low incremental contribution will be exclution from network. The result shows that the incremental contribution methods is capable reducing the size of the network is propozed, so getting optimal architecture from network.
METODE UNWEIGHTED MEANS UNTUK FAKTORIAL TAK SEIMBANG DISPROPORSIONAL WURYANDARI, TRIASTUTI
MATEMATIKA Vol 10, No 1 (2007): JURNAL MATEMATIKA
Publisher : MATEMATIKA

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (62.155 KB)

Abstract

A factorial design should be used when there are several factors of interest in experiment. The problem which arises when the factorial experiment does not contain equal sized samples in the cells is that the design is nonorthogonal. In other words, the total sums of squares in the ANOVA table cannot be decomposed into a series of additive components which permit the analysis of the separate effect. There are possible situations when the cell sample sized are unequal, allocated proportionately or disproportionately. On the allocated disproportionately, approximate F test using the method of unweighted means.