Rahmadanty, Dinda Nuranisa (2023) Peramalan Total Penjualan Produk Kosmetik Merek "E" Seri "BS" di Tokopedia Menggunakan Metode Gaussian-AR, Poisson-AR dan Binomial Negatif-AR. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 06211940000033-Undegraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
06211940000033-Undegraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2025. Download (2MB) | Request a copy |
Abstract
Industri kosmetika di Indonesia mengalami perkembangan yang cukup pesat. PT XYZ adalah salah satu perusahaan manufaktur yang tergolong ke dalam Fast Moving Consumer Goods (FMCG) bidang kosmetik yang berdiri sejak tahun 1980-an. Salah satu merek yang dinaungi perusahaan ini adalah produk kosmetik merek “E” seri “BS”. Permasalahan yang sering dihadapi adalah keterangan “Stok Habis” di toko marketplace yang tidak bisa di restock karena produk tidak tersedia di pusat distribusi nasional. Selain itu, kondisi kelebihan stok persediaan membuat produk cukup lama tersimpan di dalam gudang karena persaingan harga di marketplace, sehingga tidak layak pakai karena sudah kedaluwarsa. Perusahaan memerlukan suatu metode untuk meramalkan total penjualan untuk periode yang akan datang agar perusahaan dapat menentukan jumlah order produk yaitu dengan Gaussian-AR. Data total penjualan produk kosmetik merek “E” seri “BS” di Tokopedia merupakan data count time series sehingga digunakan metode Poisson-AR dan Binomial Negatif-AR. Ketiga metode dibandingkan performanya agar dapat menemukan model terbaik untuk peramalan berdasarkan kriteria RMSE dan MAPE. Model dengan nilai RMSE dan MAPE terkecil merupakan model terbaik. Hasil penelitian menunjukkan bahwa model Poisson-AR(1,d,[4,9,20]) merupakan model terbaik untuk meramalkan data total penjualan produk kosmetik merek “E” seri “BS” di Tokopedia dengan nilai RMSE 1.148,627 dan nilai MAPE sebesar 36,33%. Prediksi total penjualan produk kosmetik merek “E” seri “BS” dari PT XYZ di Tokopedia paling rendah terjadi pada periode minggu ke-195 Bulan Mei 2023 yaitu sebanyak 1.588 unit dan paling tinggi terjadi pada periode minggu ke-184 Bulan Maret 2023 yaitu sebanyak 4.978 unit.
===============================================================================================================================
The cosmetic industry in Indonesia has developed quite rapidly. PT XYZ is a manufacturing company that is classified in the Fast-Moving Consumer Goods (FMCG) of cosmetic industry that was founded in the 1980s. One of the brands under this company is the cosmetic products “E” brand “BS” series. The problem that is often encountered is the “Out of Stock” statement at the official store marketplace which can not be restocked because the product is not available at the national distribution center. In addition, overstock conditions cause the product to be stored in the warehouse for quite a long time due to price competition in the marketplace, consequently the product is not suitable to use because it has expired. The company needs a method to forecast total sales for the upcoming period so the company can determine the amount of ordered product that is using the Gaussian-AR. Data on total sales of the cosmetic products “E” brand “BS” series at Tokopedia is count time series data, so the Poisson-AR and Negative Binomial-AR methods are also used. The three methods are being compared to their performance in order to find the best model for forecasting based on RMSE and MAPE criteria. The model with the smallest RMSE and MAPE values is the best model. The results show that the Poisson-AR(1,d,[4,9,20]) model is the best model for forecasting data total sales of the cosmetic products “E” brand “BS” series at Tokopedia with RMSE with value of 1,148.627 and MAPE with a value of 36.33%. The prediction of total sales of the cosmetic products “E” brand “BS” series from PT XYZ at Tokopedia with the lowest sales occurs in the 195th week of May 2023 as many as 1.588 units and the highest occurs in the 184th week of March 2023 as many as 4.978 units.
Actions (login required)
 |
View Item |