Pradita, Dias Ardha (2016) Sistem Informasi Optimalisasi Perencanaan Permintaan Obat Dengan Metode Cluster Time Series Dan Neural Network Di Apotek K24 Pucang Surabaya. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
1314105065-Undergraduate_Thesis.pdf - Accepted Version Download (1MB) | Preview |
Preview |
Text
1314105065-Paper.pdf - Accepted Version Download (330kB) | Preview |
Preview |
Text
1314105065-Presentation.pdf - Presentation Download (2MB) | Preview |
Abstract
Obat merupakan komponen penting di bidang kesehatan. Salah
satu sarana kesehatan adalah apotek. Persediaan obat di apotek
harus selalu terjaga. Optimalisasi permintaan obat dapat dibantu
menggunakan metode peramalan. Namun jenis obat di apotek
sangat banyak dan tidak mungkin untuk diramalkan satu persatu,
sehingga diperlukan pengelompokan menggunakan cluster time
series untuk selanjutnya dilakukan peramalan menggunakan
neural network. Pengelompokan didasarkan pada visualisasi
dendogram menggunakan jarak autocorrelation. Identifikasi
neural network dilakukan pada tiap anggota cluster sebanyak 10
kali training dengan membandingkan antara 1 hingga 5 neuron
menggunakan metode feedforward. Tiap cluster dilakukan
peramalan dengan input, neuron dan bobot paling optimum yang
dilihat dari nilai RMSE terkecil. Data yang digunakan adalah
permintaan obat dari bulan Maret 2015 hingga April 2016. Jenis
obat yang digunakan sebanyak 7888 jenis dan di reduksi menjadi
130 jenis. Pembuatan sistem informasi diperlukan agar
kedepannya pihak apotek dapat melakukan peramalan sendiri.
====================================================================================================
Medicine is one of the most important components the health
sector. One of the medical facilities in the society is a pharmacy.
Supplies of medications in pharmacies must always be maintained
to fulfill the needs of society. Optimization of medicine demand can
be helped by using forecasting methods. But the type of medicine
at the pharmacy it very much and impossible to predict one by one,
so it requires grouping by using cluster time series. A grouping is
based on visualization dendogram and using autocorrelation
based distance. Identification of the neural network performed at
every cluster member as many as 10 times the training by
comparing between 1 to 5 neurons using feedforward method.
Every cluster performed forecasting with input neurons and the
most optimum weights were seen from the smallest RMSE value.
The data used is the demand for medicine from March 2015 until
April 2016. Kind of medicine used is 7888 types and reduced to
130 types. Information systems need to be made, so that in the
future the pharmacy can do their own forecasting.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Additional Information: | RSSt 519.55 Pra s 3100016065839 |
| Uncontrolled Keywords: | Obat, Cluster Time Series, Dendogram, Neural Network, Feedforward, Jarak Autocorrelation, Medicine, Cluster Time Series, Dendogram, Neural Network, Feedforward, Autocorrelation Based Distance |
| Subjects: | H Social Sciences > HA Statistics > HA30.3 Time-series analysis |
| Divisions: | Faculty of Mathematics, Computation, and Data Science > Statistics > 49201-(S1) Undergraduate Thesis |
| Depositing User: | - Davi Wah |
| Date Deposited: | 25 Feb 2020 07:41 |
| Last Modified: | 25 Feb 2020 07:41 |
| URI: | http://repository.its.ac.id/id/eprint/75120 |
Actions (login required)
 |
View Item |