網際網路的興起為傳統的投資交易模式帶來革命性的衝擊，國內外證券商紛紛加入網路交易的行列，因此針對網路顧客的需求來瞭解顧客所重視的服務品質項目以提昇服務品質，儼然已成為重要的課題。本研究嘗試將類神經網路辨試類別的特性應用於證券業網路下單執行服務品質的類別分析，並利用多變量分析中的因素分析將顧客對服務品質重要項目的滿意度化成數項構面，再將這幾個構面交予類神經網路學習來建立網路模式，以提供證券業作為執行服務品質管理成效評家的參考依據。本研究所構建的網路模型分類歸屬能力是以辨識率為評估指標，綜合類神經網路與統計學之區別分析的研究，實證結果顯示類神經網路在學習樣本或測試樣本上，辦識率分別高達99.82%與99.65%，而傳統的區別分析其整體辨識率只有99.33%。因此，若採用類神經網路進行分類判斷，有較高的辨識率，能產生較好的分類結果，所以運用類神經網路於網路下單執行服務品質管理分類，其分類效果及穩定度皆比區別分析為佳，是一項值得建議使用的工具。

With the development of internet financial investment, domestic and foreign securities have scrambled to launch online stock. To focus on customer demands of online trading investors, understand the main factors that customers really care, and improve their service quality are becoming increasingly important issues. This study applies the classification characteristics of artificial neural network to the categorizing level of service quality for online trading stock companies. Back-Propagation neural network is adopted as the fundamental framework for this study. The objective of this study is to extract several dimensions of the service quality of the customers’ satisfaction through the factor analysis of the multivariate analysis. These dimensions are taken as the input variables to build a neural network model through the learning process. The results validate that the classification performance of neural network is better than discriminant analysis. Neural network is very useful tool for online trading stock companies to execute the performance measurement when they implement service quality activities.