sDNA 4.1.1英文文档学习系列 3. 分析：结果的含义是什么？-优快云博客

本文链接：https://blog.youkuaiyun.com/m0_52564979/article/details/144343749

写在最前面

其实读完第二章的内容云里雾里的。继续第三章的内容。在第三章正文开始前，用户手册强烈建议我们先准备好我们的网络然后继续进入后面的内容~~因为我本专业不是学习这个方面的，很多单词我不太确定他的准确中文翻译是什么，我就直接用英文替代了。原文其实很长有很多描述性的部分和帮助理解的部分。

正文部分

3.1. Links and network radius¶

首先是关于network link的定义：注意junction和dead end。这里有一个实际网络中直线街道的例子来让我们理解相关的概念。

“ link density is a good measure of urban density”这一章的主图是 Cardiff link density, 1km radius.通过这个图片（一图胜千言，(*^_^*)）我们可以更好的理解实际的概念。

这里的density和radius是如何扯上关系的？实际上，图中的密度是基于每个link的计数，即在其周围1000米半径内（不是圆的那种理解，继续往后看）有多少个link的存在。 Links are the fundamental unit from which spatial networks are built, and for that reason it makes sense to standardize on them as a unit of description. Also, link density tends to hold information about the network. For example, in a road network, link density can correlate over 90% with the density of houses and jobs.由此可见，第二章中提到的网络准备是很重要的，合理的数据才能得到合适的结果。

R1000 – the 1000m radius的含义：all points that can be reached by travelling 1000m through the network from that link。从该link出发通过网络行驶 1000m 可以到达的所有点。It’s definitely not a circle.原文给了一个图片的示例来帮助我们理解这个问题，非常直观。

此外在分析结果中通常包含了Rxxx/Rn或者Rxxxxc/Rnc，他们都有自己的含义。c表示是连续空间的分析。Continuous space analysis treats links as continuous entities; if part of the link falls inside the radius, and part outside, then the link is split on-the-fly and we count only the part of that falls inside. Continuous space should always give the most accurate results, but discrete space is faster to compute

sDNA提供的其他量化网络的方法：

Total Length is self explanatory
Number of junctions is self explanatory
Total Weight is the total weight customizable by the user
Connectivity is the total number of link ends connected at junctions

3.2. Betweenness

The flow model inherent in sDNA is based on Betweenness.

Betweenness分析假设网络由从任何地方到任何其他地方的实体组成，并且这些实体的最大行程距离由半径确定。我们假设这些实体通过最短路径旅行——我们称这种路径为地理测地线（geodesic）

Betweenness analysis assumes your network is populated with entities that go from everywhere to everywhere else, subject to a maximum trip distance determined by the radius. We assume these entities travel via the shortest possible path – and we call such a path a geodesic. But how we define “shortest” may vary. We call the definition of distance a metric; currently sDNA supports

Euclidean metrics, taking the shortest physical distance possible
angular metrics, which minimize the amount of turning both on links and at junctions 角度度量的含义感觉有些抽象，但是通过手册中实际的例子可能比较好理解：“second on the right then straight on ‘till morning” ； Drivers of vehicles also tend to follow angular geodesics, but for a different reason – straight roads through a city tend to be faster, on average.
custom metrics based on user data
hybrid metrics which are user programmable
specialist metrics for cyclists, vehicles and pedestrians

A final question in betweenness analysis is what weight to assign to each geodesic. 如何为每个测地线分配权重？分配什么样的权重？

The simplest answer is to assign a weight of 1 to every pair of links. As the density of jobs and homes is strongly related to the density of network links, betweenness weighted in this manner usually correlates well with traffic flows. But what if network length is important to you? Let’s say you want to assume longer streets act as origins and destinations for more people, or you’re analyzing rivers, etc. Or maybe you have custom weights - census data for example, or building entrances. In this case we assume the number of entities making the trip is proportional both to the weight of the origin and the weight of the destination, so we multiply the origin weight and destination weight together to compute the betweenness weight.这段其实有点抽象，我没太读懂，具体实操的时候再体会一下吧。

3.3. Closeness

Closeness中心性度量与Betweenness类似，也是网络中心性的一种形式，它与常见的可达性（accessibility）的概念相吻合。

平均距离（Mean Distance）最常见的Closeness度量。 It measures the difficulty, on average, of navigating to all possible destinations in radius x from each link. Technically then it’s a form of farness not closeness: this only means that big numbers mean “far” instead of “close”.

sDNA names its closeness outputs Mean Angular Distance (MAD), Mean Euclidean Distance (MED), Mean Custom Distance (MCD), etc.

Note that the “mean” of mean distance is weighted by the weight of destinations.