output	editor_options
revealjs::revealjs_presentation self_contained transition theme reveal_options css true fade simple controls slideNumber margin width height false true 0.05 1024 780 slides.css	chunk_output_type console

netplot: Beautiful graph drawing

• An alternative graph visualization engine that puts an emphasis on aesthetics at the same time of providing default parameters that provide visualizations that are out-of-the-box nice.

Some features:

• Auto-scaling of vertices using sizes relative to the plotting device.
• Embedded edge color mixer.
• True curved edges drawing.
• User-defined edge curvature.
• Nicer vertex frame color.
• Better use of space filling the plotting device.

The package uses the grid plotting system (just like ggplot2).

Comparison

{style="width:600px"}

UK Faculty

{style="width:600px"}

Some features

Node scaling

{style="height:600px"}

Some features

Node shapes

{style="height:600px"}

Some features

Edge curvature

{style="width:800px"}

Some features

Edge type of line

{style="height:600px"}

US airports

{style="height:600px"}

Applied Social Network Analysis with R

Little ERGMs

The distribution of $\mathbf{Y}$ can be parameterized in the form

$$ \Pr\left(\mathbf{Y}=\mathbf{y}|\theta, \mathcal{Y}\right) = \frac{\exp{\theta^{\mbox{T}}\mathbf{g}(\mathbf{y})}}{\kappa\left(\theta, \mathcal{Y}\right)},\quad\mathbf{y}\in\mathcal{Y} \tag{1} $$

Where $\theta\in\Omega\subset\mathbb{R}^q$ is the vector of model coefficients and $\mathbf{g}(\mathbf{y})$ is a q-vector of statistics based on the adjacency matrix $\mathbf{y}$.

---

• Model (1) may be expanded by replacing $\mathbf{g}(\mathbf{y})$ with $\mathbf{g}(\mathbf{y}, \mathbf{X})$ to allow for additional covariate information $\mathbf{X}$ about the network. The denominator,

  $$ \kappa\left(\theta,\mathcal{Y}\right) = \sum_{\mathbf{z}\in\mathcal{Y}}\exp{\theta^{\mbox{T}}\mathbf{g}(\mathbf{z})} $$

• Is the normalizing factor that ensures that equation (1) is a legitimate probability distribution.

• Even after fixing $\mathcal{Y}$ to be all the networks that have size $n$, the size of $\mathcal{Y}$ makes this type of models hard to estimate as there are $N = 2^{n(n-1)}$ possible networks!

The lergm R package

• An Extension of the ergm (regular size fitting via simulation) package

• Uses exact statistics for fitting small networks (3 to 6 nodes).

• Will be designed mostly to be ran with multiple networks simulatenously (so we recover the asymptotic properties of the MLE estimators)

• Work in progress...

Thanks!

• netplot: https://github.com/USCCANA/netplot

• Applied SNA with R: https://gvegayon.github.io/appliedsnar/

• Little ERGMs: https://github.com/USCCANA/social-smarts/

Twitter: @gvegayon

email: vegayon@usc.edu