| dc.description.abstract | In this thesis, there is an implementation of an ecommerce platform case study solved with
thousands of orders per day. However, orders are controlled by human and there is no involment
of techniques to handle order processing tasks. Due to this difficulty, this thesis developed a
clutering order technique by machine learning to be considered as solution for this company.
First, data is gathered and processed to be used for the whole report. Through numerous features,
weight, location, and price are three important features for the analysis and gone throught the
entire of this study. After feature chosen, Kmeans and Fuzzy c means are compared to choose the
best one for the model of this study. To combine all clusters of each feature, between decision
tree and XG boosting, decision tree with scoring boosting is the effective way with high accuray
to gain the high performance. Besides, the flow of this study is also compared with benchmark to
find the optimal way for the model. | en_US |
