| dc.contributor | 經濟系 | |
| dc.creator (作者) | 陳心蘋 | |
| dc.date (日期) | 2020-10 | |
| dc.date.accessioned | 16-Jul-2025 11:12:33 (UTC+8) | - |
| dc.date.available | 16-Jul-2025 11:12:33 (UTC+8) | - |
| dc.date.issued (上傳時間) | 16-Jul-2025 11:12:33 (UTC+8) | - |
| dc.identifier.uri (URI) | https://ah.lib.nccu.edu.tw/item?item_id=177594 | - |
| dc.description.abstract (摘要) | 空間資訊加上時間變數的分析有助於區域相關資料的分析。空間分析模型中的空間權數矩陣衡量空間單位間的互動程度，處理觀察值空間相依的特質：空間模型包含內生互動效果，外生互動效果，與誤差項間的互動效果。傳統的空間模型假設空間權數矩陣內的參數為外生且固定的，然而區域內的空間相依互動程度，可能隨區域或時間而改變，如網路空間資訊。時間分析模型中，觀察值的時間相依特性，通常以時間序列模型中的自我回歸與移動平均模型處理。相關研究曾結合空間與時間資料分析，然文獻中的空間權數矩陣內的參數為固定的，且時空分析模型中缺乏空間的外生相依效果，即解釋變數的空間互動項。許多的動態空間資料如區域與交通資料具有網絡特質，但相關的空間時間資料分析法中，很少納入網絡特性探討時空資料，也欠缺外生相依效果和以網絡結構為基礎的模型。傳統的網絡系統中，以每個節點和其他節點的連結總數衡量該節點的連結程度，然而在實際網絡中，許多網絡內節點間的連結不是單一選擇，而是有程度之分的。如交通網絡中，城市與城市間的交通成本因距離或路況而不同。加權網絡系統容許節點間連結程度不同。本研究擬結合空間分析與時間序列模型，推導出以網絡系統為基礎的時間空間分析模型；分析模型中空間、時間之相依效果與網絡效果的檢定；空間權數矩陣的模式與特性分析。 | |
| dc.description.abstract (摘要) | There is a common statistical characteristic for spatio-temporal data that nearby (space and time) units tend to be more interacted than those far apart described as spatial and temporal dependence. The spatial dependence characteristic is captured by the spatial weight matrix modified in spatial analysis, which allows for endogenous interaction effect, interaction effects among the error terms and exogenous interaction effects. The temporal dependence characteristic is commonly presented by time series analysis such as autoregressive and moving average models. The space-time autoregressive moving average and its various extensions have been proposed for describing spatio-temporal processes. The parameters denoting the spatial and temporal correlations in these models are assumed to be fixed globally, which is inadequate to describe data with dynamic and heterogeneous dependence. There is lack of exogenous spatial interaction effects in these models. Many dynamic spatial data in regional or transportation analysis contains network structure. In a traditional network system, the degree of each node is the sum of the links to all other nodes. Each link weights the same. However, the connections among nodes in many real-world spatial networks are not merely binary choice, but rather have different strength. The spatial weights which reflect the strength of the interaction in spatial model characterize the corresponding network. The degree of interaction for each pair of nodes could be changed by time and location. The form and the feature of the spatial weights matrix affects the generating process of the network and the resulting size distribution of the region. The dynamic spatial network system consists of spatio-temporal data; there is lack of research analyzing the spatio-temporal data from the network point of view. The propose of this work is to modify a spatio-temporal model based on a dynamic spatial network system, to investigate appropriate parameterization and the role of each parameter in the model, to develop the corresponding hypothesis testing for spatial or temporal dependences and network effects, to examine the form and estimation method of the corresponding spatial weight matrix, and to empirically examine the feature of the model by calibrating the parameter and simulating the model. | |
| dc.format.extent | 116 bytes | - |
| dc.format.mimetype | text/html | - |
| dc.relation (關聯) | 科技部, MOST107-2119-M004-001, 107.08-108.07 | |
| dc.subject (關鍵詞) | 空間權數矩陣; 加權網絡 | |
| dc.subject (關鍵詞) | Spatial weight matrix; Weighted network | |
| dc.title (題名) | 網絡的空間時間資訊分析 | |
| dc.title (題名) | The Analysis of the Spatio-Temporal Data of Network | |
| dc.type (資料類型) | report | |
