查詢結果分析
來源資料
頁籤選單縮合
題 名 | 應用資料探勘技術探討客戶申裝電信服務其頻繁型樣=Using Data Mining Technique to Explore the Frequent Patterns on Telecom Business Customer's Service Data |
---|---|
作 者 | 傅瑞玫; 許清琦; | 書刊名 | 電信研究 |
卷 期 | 36:4 民95.08 |
頁 次 | 頁471-496 |
分類號 | 557.7 |
關鍵詞 | 資料探勘; 關聯法則; 頻繁型樣; 頻繁型樣樹; 頻繁型樣增生; |
語 文 | 中文(Chinese) |
中文摘要 | 使用資料探勘(Data Mining)之相關技術以探討電信產業訂單資料庫中客戶電信服務的特性是一項值得探討的研究,引用 "Mining Frequent Patterns without Candidate Generation" 之理論,以 FP-tree 資料架構來構建客戶服務的資料,藉由 FP-Growth 演算法產生頻繁型樣,中華電信訂單資料的頻繁型樣是否存在,是否能夠藉此演算法之實作產生?若確實產出是否有其價值?同時藉由實作提出對此演算法的幾項改進,讓此演算法更能應用於所被探勘的中華電信訂單資料及提昇演算法的重要性及實用價值,並將所探勘出來之頻繁型樣可以彈性動態地予以調整,為滿足研究宗旨因此將客戶資料分類化、動態給定minimum support ,動態給定各分類的minimum support 等改善方法,期望藉由這些改善方法及程序能夠適當及有效地將 FP-tree及FP-Growth 演算法所產生過多的頻繁型樣適當修剪,並以DGFP(Dynamically Generate Frequent Patterns),一個互動模式動態產生頻繁型樣的工具,以方便使用者進行資料探勘。 |
英文摘要 | Using data mining technique to explore the customer service characteristics from our order processing database is an issue worth studying. This paper uses the "Mining Frequent Patterns without Candidate Generation" algorithm and uses FP-tree structure to construct customer frequent pattern data tree, and uses FP-growth method to generate the frequent patterns to see whether the frequent patterns can be discovered and are valuable. Also by making several improvements on the approaches, we are able to broaden and multiply the basic algorithm value, such as classifying the customer service data, applying constraints and different thresholds for each class to prune frequent patterns properly and effectively, and creating a DFPG ( Dynamically Generate Frequent Patterns ) to let users interactively online generate frequent patterns dynamically. |
本系統中英文摘要資訊取自各篇刊載內容。