SFUSD has released the Round 1 Demand data for the 2010-2011 school year. The data is in PDF format, making it rather hard to do any analysis. For the benefit for myself and others I've converted the K-level data to spreadsheet format. For example, I used the data to generate the over/under-subscription chart below. I subtracted the number of first choice requests from the capacity to come up with the absolute over-subscription number.

It also shows that the top 10 school programs account for 72% of total number of over-subscription. Two third of the programs are balanced or under subscripted with respect to the number first choice. This includes the popular West Portal GE! It seems not that hard to find some good school with good confidence to get in. Of course the Diversity Index system makes everything unpredictable.

2010.03.15 [San Francisco] - comments

The Peak Traffic puzzle is easy to interpret. It is a graph problem to find all maximal cliques. Implementating it, like many graph problems, is quite difficult though. The possible combination is huge. And there is no simple ordering graph subset. It end up costed me several days to do it. Here is the outline of my solution:

1. Parse the input. Generate an undirected graph.
2. Generate a list of cliques using recursive algorithm.
3. Eliminate subsets from the result to find all maximal cliques.

After working on it for several days, I'm quite done. I haven't spent any more effort to optimize it and expain it. So here is my un-tidied code.

2010.03.03 [programming] - comments