Publications-Theses

Article View/Open

Publication Export

Google ScholarTM

NCCU Library

Citation Infomation

Related Publications in TAIR

題名 大數據分析電子公司維修狀況
Big data analysis for maintenance status of electronics company
作者 郭九一
Kuo, Chou-Yi
貢獻者 周珮婷
Chou, Pei-Ting
郭九一
Kuo,Chou-Yi
關鍵詞 存活分析
類神經網路
探索性資料分析
非監督式學習
Survival analysis
Unsupervised learning
Exploratory data analysis
Neural networks
日期 2018
上傳時間 3-Jul-2018 17:24:01 (UTC+8)
摘要 本研究主要探討以A公司為例,產品可能會受到天氣影響熱漲冷縮造成之維修因素,因此探討產品維修量在中國地區是否受季節因素影響,並準確預測其產品每月之維修數量。首先因為中國維修站以及維修點位種類眾多,我們以非監度式學習的方法,將中國大陸的各維修站在切割式分群下,利用溫度搭配經緯度分群,同樣的將維修點位在切割式分群下,利用每個月的維修數量分群,因某些點位之間數量差異甚大以致於群間差異大,接著進行資料探索,找出季節對於不同特殊維修點位的維修數量是否存在顯著影響,以及找出分群過後的維修站在不同維修點位下的分配,利用視覺化的方式觀察出夏天對於維修數量有顯著影響,且越接近沿海地帶與緯度低者最為顯著。接著利用類神經網路預測在維修點位數量最多的LP01上的產品維修數量,分別適配兩種類神經網路,一種是倒傳遞網路(Backpropagation Network),有接收、傳遞、產生等基本功能,其中包含處理單元、層與網路,因隱藏層的關係,它們允許變數與預測變量之間的非線性關係,透過已知變數輸入到輸入層,直接預測維修數量與維修率,另一種方法為時間序列類神經網路(Neural network autoregression),預測效果比倒傳遞網路好,可以有效預測在不同維修站群在維修產品上的分配與走向。
The current study investigated the reason for product repairs in the Enterprise A.
The impact of weather on the amount of repaired product was discussed and the amount of future product repair was predicted. First, unsupervised learning methods were used to group the repair locations in China by their temperature, latitude, and longitude. In addition, repair parts were grouped using the number of repairs per month. Second, EDA was used to discuss the charateristics of each group of repaired parts. Later, Neural network techniques were used to predict the amount of future product repairs on LP01. The results of back-propagation network and neural network autoregression were compared. We found that sever weather affect the amount of product repairs. Hot temperature can significantly impact the performance of product, especially in coastal areas and low latitude regions. We suggest the company to establish a standard way or rule to collect and store maintenance and prouct repair information for future analysis.
參考文獻 Abbas, O. A. (2008). Comparisons Between Data Clustering Algorithms. Int. Arab J.
Inf. Technol., 5(3), 320-325.
Athana¬sopou¬los, G. &. Hyndman, R. J. (2014).Forecasting: principlesand practice. Retrieved from https://www.otexts.org/fpp/9/3
Boukelif, A. & Faraoun, K. M. (2005). Neural Networks Learning Improvement using the K-Means Clustering Algorithm to Detect Network Intrusions
Ferreira, L. & Hitchcock, D. B.(2009). A Comparison of Hierarchical Methods for Clustering Functional Data doi: 10.1080/03610910903168603
Murtagh, F., & Legendre, P. (2014). Ward’s Hierarchical Agglomerative Clustering Method:Which Algorithms Implement Ward’s Criterion Journal of Classification 31:274-295 doi: 10.1007/s00357-014-9161-z
高千琇(民93)。《工業區設置對台灣地區製造業廠商存活之影響—以電力及電子機械器材製造修配為例》
陳翠玲、紀雍華、陳孟詩、林沛練(民104)。應用縱向資料K-means群集方法之臺灣雨量分區研究。
描述 碩士
國立政治大學
統計學系
105354017
資料來源 http://thesis.lib.nccu.edu.tw/record/#G1053540171
資料類型 thesis
dc.contributor.advisor 周珮婷zh_TW
dc.contributor.advisor Chou, Pei-Tingen_US
dc.contributor.author (Authors) 郭九一zh_TW
dc.contributor.author (Authors) Kuo,Chou-Yien_US
dc.creator (作者) 郭九一zh_TW
dc.creator (作者) Kuo, Chou-Yien_US
dc.date (日期) 2018en_US
dc.date.accessioned 3-Jul-2018 17:24:01 (UTC+8)-
dc.date.available 3-Jul-2018 17:24:01 (UTC+8)-
dc.date.issued (上傳時間) 3-Jul-2018 17:24:01 (UTC+8)-
dc.identifier (Other Identifiers) G1053540171en_US
dc.identifier.uri (URI) http://nccur.lib.nccu.edu.tw/handle/140.119/118222-
dc.description (描述) 碩士zh_TW
dc.description (描述) 國立政治大學zh_TW
dc.description (描述) 統計學系zh_TW
dc.description (描述) 105354017zh_TW
dc.description.abstract (摘要) 本研究主要探討以A公司為例,產品可能會受到天氣影響熱漲冷縮造成之維修因素,因此探討產品維修量在中國地區是否受季節因素影響,並準確預測其產品每月之維修數量。首先因為中國維修站以及維修點位種類眾多,我們以非監度式學習的方法,將中國大陸的各維修站在切割式分群下,利用溫度搭配經緯度分群,同樣的將維修點位在切割式分群下,利用每個月的維修數量分群,因某些點位之間數量差異甚大以致於群間差異大,接著進行資料探索,找出季節對於不同特殊維修點位的維修數量是否存在顯著影響,以及找出分群過後的維修站在不同維修點位下的分配,利用視覺化的方式觀察出夏天對於維修數量有顯著影響,且越接近沿海地帶與緯度低者最為顯著。接著利用類神經網路預測在維修點位數量最多的LP01上的產品維修數量,分別適配兩種類神經網路,一種是倒傳遞網路(Backpropagation Network),有接收、傳遞、產生等基本功能,其中包含處理單元、層與網路,因隱藏層的關係,它們允許變數與預測變量之間的非線性關係,透過已知變數輸入到輸入層,直接預測維修數量與維修率,另一種方法為時間序列類神經網路(Neural network autoregression),預測效果比倒傳遞網路好,可以有效預測在不同維修站群在維修產品上的分配與走向。zh_TW
dc.description.abstract (摘要) The current study investigated the reason for product repairs in the Enterprise A.
The impact of weather on the amount of repaired product was discussed and the amount of future product repair was predicted. First, unsupervised learning methods were used to group the repair locations in China by their temperature, latitude, and longitude. In addition, repair parts were grouped using the number of repairs per month. Second, EDA was used to discuss the charateristics of each group of repaired parts. Later, Neural network techniques were used to predict the amount of future product repairs on LP01. The results of back-propagation network and neural network autoregression were compared. We found that sever weather affect the amount of product repairs. Hot temperature can significantly impact the performance of product, especially in coastal areas and low latitude regions. We suggest the company to establish a standard way or rule to collect and store maintenance and prouct repair information for future analysis.
en_US
dc.description.tableofcontents 第一章 緒論…………………………………………1
第二章 文獻探討……………………………………2
第三章 資料敘述……………………………………4
第四章 研究方法……………………………………6
第一節 非監督式學習………………………………6
第二節 存活分析……………………………………8
第三節 時間序列……………………………………9
第五章 研究過程與結果……………………………12
第一節 資料分群……………………………………12
第二節 資料探索……………………………………26
第三節 存活分析 Kaplan-Meier……………………31
第四節 模型-類神經網路……………………………35
第六章 結論…………………………………………50
第一節 研究結果……………………………………50
第二節 未來發展與建議……………………………51
第七章 參考資料………………………………………53
zh_TW
dc.format.extent 2822125 bytes-
dc.format.mimetype application/pdf-
dc.source.uri (資料來源) http://thesis.lib.nccu.edu.tw/record/#G1053540171en_US
dc.subject (關鍵詞) 存活分析zh_TW
dc.subject (關鍵詞) 類神經網路zh_TW
dc.subject (關鍵詞) 探索性資料分析zh_TW
dc.subject (關鍵詞) 非監督式學習zh_TW
dc.subject (關鍵詞) Survival analysisen_US
dc.subject (關鍵詞) Unsupervised learningen_US
dc.subject (關鍵詞) Exploratory data analysisen_US
dc.subject (關鍵詞) Neural networksen_US
dc.title (題名) 大數據分析電子公司維修狀況zh_TW
dc.title (題名) Big data analysis for maintenance status of electronics companyen_US
dc.type (資料類型) thesisen_US
dc.relation.reference (參考文獻) Abbas, O. A. (2008). Comparisons Between Data Clustering Algorithms. Int. Arab J.
Inf. Technol., 5(3), 320-325.
Athana¬sopou¬los, G. &. Hyndman, R. J. (2014).Forecasting: principlesand practice. Retrieved from https://www.otexts.org/fpp/9/3
Boukelif, A. & Faraoun, K. M. (2005). Neural Networks Learning Improvement using the K-Means Clustering Algorithm to Detect Network Intrusions
Ferreira, L. & Hitchcock, D. B.(2009). A Comparison of Hierarchical Methods for Clustering Functional Data doi: 10.1080/03610910903168603
Murtagh, F., & Legendre, P. (2014). Ward’s Hierarchical Agglomerative Clustering Method:Which Algorithms Implement Ward’s Criterion Journal of Classification 31:274-295 doi: 10.1007/s00357-014-9161-z
高千琇(民93)。《工業區設置對台灣地區製造業廠商存活之影響—以電力及電子機械器材製造修配為例》
陳翠玲、紀雍華、陳孟詩、林沛練(民104)。應用縱向資料K-means群集方法之臺灣雨量分區研究。
zh_TW
dc.identifier.doi (DOI) 10.6814/THE.NCCU.STAT.001.2018.B03-