完整後設資料紀錄
DC 欄位 | 值 | 語言 |
---|---|---|
dc.contributor.author | 張勝雄 | en_US |
dc.contributor.author | 劉敦仁 | en_US |
dc.date.accessioned | 2014-12-12T02:40:39Z | - |
dc.date.available | 2014-12-12T02:40:39Z | - |
dc.date.issued | 2013 | en_US |
dc.identifier.uri | http://140.113.39.130/cdrfb3/record/nctu/#GT070163426 | en_US |
dc.identifier.uri | http://hdl.handle.net/11536/74467 | - |
dc.description.abstract | 在網路資訊快速成長的時代,隨著使用的載具多元化,消費者在網路留下的瀏覽記錄數量也是呈現大幅度的成長,在大量的資料記錄累積下,尋找執行效能比較合適的解決方案,分析工具的選擇也成為一個重要的議題。 目前市面上知名的開放原始碼分析工具有R與Mahout,它們皆能搭配Hadoop平台進行運作。本研究主要是透過實驗的方式來了解,兩工具在推薦演算法執行效能上的差異。研究結果發現,以wordcount之實驗,使用Hadoop-streaming搭配R編寫之程式進行測試所得到的執行效能與使用Hadoop native程式的執行效能差不多;但在有矩陣運算時,Mahout與R兩者在記憶體的支配上就出現明顯差異。本研究對於Mahout及R兩者的執行效能測試結果分析,可提供欲使用Hadoop搭配Mahout及R進行運算之使用者參考。 | zh_TW |
dc.description.abstract | In recent years, the number of browsing tools, customers, and the amount of data have grown exponentially. How to select data analysis tool is an important issue for decision-maker. The two popular open sources of analysis tool are R and Mahout. Both of them can operate on Hadoop platform. The purpose of this research is to understand the recommendation algorithm operation performance with R and Mahout. By conducting several experiments. The experiment results of wordcount data analysis show that Hadoop-streaming with R can get the same performance with Hadoop native. For the experiments involving matrix calculations, R requires larger memory capacity than Mahout. | en_US |
dc.language.iso | zh_TW | en_US |
dc.subject | 分散式 | zh_TW |
dc.subject | 效能比較 | zh_TW |
dc.subject | Distribution | en_US |
dc.subject | Performance | en_US |
dc.title | 基於Hadoop平台之物品推薦效能比較分析 | zh_TW |
dc.title | Comparative Analysis of the Performance of Item Recommendations based on Hadoop Platforms | en_US |
dc.type | Thesis | en_US |
dc.contributor.department | 管理學院資訊管理學程 | zh_TW |
顯示於類別: | 畢業論文 |