criminal_cascades - [no description]

Age	Commit message (Collapse)	Author
2015-07-01	Got predict-victims running in parallel, drastically reducing the time	Ben Green
	for each test. Also changed how we get the rankings of infected individuals each day.
2015-06-28	Worked on synthetic data recovery so we can tell how high the actual	Ben Green
	infector is ranked among all potential parents. Cleaned up code for the predicting victims benchmarking test.
2015-06-25	Worked on process to generate cascades in R, recover them in ml2, and	Ben Green
	compare true/recovered. Results look good: ml2 finds close-to-actual parameters and gets ~60% of the infectors right. I will check how many recovery gets in the top 2 or 3 infectors.
2015-06-25	Tried ml3 that allows for multiple parents to infect.	Ben Green

2015-06-25	playing around with ml3	Ben Green

2015-06-23	ran some experiments with ml3, still no good results	Ben Green

2015-06-23	updated calculation of structural component with edge weights, and to	Ben Green
	penalize less for longer distances
2015-06-22	Added ml3, where we don’t consider edges from victims to non-victims.	Ben Green
	Also playing around with only allowing most likely parent to try and infect. Still working to see if some variant produces good results that allow us to determine seeds/non-seeds.
2015-06-20	altered weight_success to show dist and dt between each parent and child	Ben Green

2015-06-20	More tests with ml2 to find optimal results	Ben Green

2015-06-20	added new model, ml2, which assumes all victims are infected by someone	Ben Green
	else (unless they are root victims)
2015-06-20	tried generating cascades and recovering parameters, still got the same	Ben Green
	result where optimization minimizes alpha, delta as much as possible
2015-06-19	Continuing to experiment. Identifying infections vs seeds might be	Ben Green
	impossible :/. We may want to try a model where we just assume infections — can get parameters for the benchmark test.
2015-06-18	testing new model, realizing we may be getting optimal beta wrong	Ben Green

2015-06-18	altered spawning model and changed temporal probability on	Ben Green
	weight_success
2015-06-16	Changed model for spawning. Began fixing underflow error in	Ben Green
	prob_success/failure
2015-06-15	minor changes for testing	Ben Green

2015-06-11	started adding weights into edge probabilities	Ben Green

2015-06-11	cleaned up code with some minor changes	Ben Green

2015-06-08	Ben’s first commit after getting code working!	Ben Green

2015-06-07	Project cleanup before handing it to Ben	Thibaut Horel

2015-04-06	Updated model, still getting weird results though	Thibaut Horel

2015-03-30	New version of the model	Thibaut Horel