Visualisation of Solution and Real Example

[AlexGibberd]

One thing we will need for any paper is an example of the recovered graph solutions on some real data.

We may consider to borrow some of the visualisation tools from the sklearn example:
http://scikit-learn.org/stable/auto_examples/applications/plot_stock_market.html#sphx-glr-auto-examples-applications-plot-stock-market-py

I'm not sure if the stock price data is too long (too many data-points) to calculate fast solutions. We can look at other data if you want, or just try it on the finance data? I have some aggregate stock market data, which I used in the estimating dynamic graphical models review paper (in the dropbox folder), see Figure 4.

Additionally, in our example, we should ideally have some way of selecting a good set of (lambda1,lambda2). This should be achieved by solving the other issues raised, but it would be useful to have some nice visualisations of the graph's that we estimate.

I think for this we could also use similar clustering/embedding techniques as from sklearn to visualise the graphs.

[AlexGibberd]

• Write a function to view individual edges in graph over time
• Visualise F-score over a grid of regulariser values
• Visualise changepoint recovery over grid of regulariser values