Logo image
Sales Forecasting for New Items with Short Histories: Comparing Strategies for Time­-Series Matching and Forecasting using Time­-Series Clusters
Thesis

Sales Forecasting for New Items with Short Histories: Comparing Strategies for Time­-Series Matching and Forecasting using Time­-Series Clusters

Lee, Tung-Yu
Masters, 國立清華大學, 服務科學研究所
2016

Abstract

銷售預測 少量的歷史資料 時間序列群集分析 原型 時間序列配對分法 Sales Forecasting Short­-history data Time­-series Clustering Prototypes Time­-series Matching
Time­-series forecasting is a popular technique and is used for different purposes in many fields. However, one big problem exists in many common forecasting algorithms is the short­-history problem. Our study presents a 3­-step method that can automatically provide forecasts for items with very short histories. This 3­-step method composes of time­-series clustering, time­-series matching, and forecasting methods. There are several different strategies in each step, so 12 combinations are conducted in our study. We compare the forecasting performance of different combinations using real sales data from iCHEF, a Taiwanese platform company. The results show that time­-series matching methods influence forecasting performance significantly. Besides, similar historical data is useful information while forecasting for items with short histories.

Metrics

1 Record Views

Details

Logo image