隨著資訊網路的興起，網路為人與人之間訊息的交流帶來了革命性的變化，網路交易變的日益頻繁，投資人也逐漸重視起資訊網路所帶來的影響。過去有許多研究透過社會網路分析的方式來探討是否會對投資人的決策造成影響，甚至是價格上的變動，網路所帶來的這些變化無疑對整個金融市場帶來了很大的影響。
本研究希望藉此探討距離在地理網路中，其券商分點與買賣單失衡的影響，本研究以集群分析的方式將數百個證券方分點以及券商公司分點做歸類整理，將同性質亦或是有類似的交易行為者歸類為一群。以集群分析而言，皆是以個別資料點與集群中心的距離計算來找出與該中心最接近的資料點，再依照演算法，將眾多的資料彙整找出分群。因此彙整完後的資料是否符合預期，就是選擇的集群個數是否為最合適的。根據實證結果發現，距離公司總部愈近的券商，因為對位於附近的公司較為熟悉，因此存在熟悉度的偏好，其結果表明比起資訊上的優勢，更多數是出自對於這家公司熟悉的偏好。
With the rise of the information network, there has been a revolutionary change in
the exchange of information among individuals. Online transactions have become increasingly frequent, and investors are gradually paying more attention to the impact of the information network. Previous research has explored whether social network analysis could influence investors' decision-making or even lead to price fluctuations. The changes brought about by the network undoubtedly have a significant impact on the entire financial market.
This study aims to investigate the impact of geographical network distances on the imbalance among brokerage branches and their buy and sell orders. I classify and organize hundreds of brokerage branches by using cluster analysis, grouping together those with similar characteristics or trading behaviors. Cluster analysis relies on measuring the distance between individual data points and the cluster center to identify the data points closest to that center. Subsequently, based on algorithms, numerous data points are aggregated to determine clustering. Thus, verifying whether the aggregated data meets expectations involves determining the optimal number of clusters to select. According to the empirical findings, it was observed that brokerages located closer to the headquarters of a company exhibit a preference for familiarity due to their proximity, suggesting a bias towards companies with which they are more acquainted.
