alanctgardner3

There are companies using Kinect to track users travelling through retail stores. There's also using existing WiFi recieved signal strength, which is not terribly granular but benefits from existing infrastructure.

Is the goal that the user will enter a binding contract to buy without knowing what they'll pay?

There already is price competition. Anyone who can Google for 5 minutes can find the lowest price, and that's probably fulfilled by Amazon or another large retailer. The idea that you'll carry every product, offer a lower price than Amazon, and convince people to participate in this weird pricing scheme is pretty absurd.

It depends on how much you want to idealize things. If you add the assumption that all flights/trains/etc. are available every day, and the price never changes, then yes. In reality, trying to traverse a weighted graph optimally with a bunch of constraints (arrival times, how long to stay in each place, preference for difference modes of transit, transit times) is a whole field of research in math/CS. Operations research [1] is the study of optimizing processes with many interrelated factors.

1. http://en.wikipedia.org/wiki/Operations_research

Did you just describe git? Git has revisions of files, and they can be tagged, merged, etc. It still works on top of a conventional file system though, because why reinvent the wheel?

Frankly, Canada has a fetish for Waterloo. I think you have some misconceptions though: their placement rate for co-ops is like 98%. Students aren't dropping out because they can't find a position, although they are forced to travel for their placements in some cases. Most of the students I know who selected UW wanted to end up at a Google, LinkedIn, Facebook, or start a company - academia is definitely not the emphasis, and I think students know that going in.

uOttawa grad here: is there any co-op option at Carleton? It wasn't very popular at uOttawa, to the point that my employers were getting Waterloo grads because they couldn't find anyone local. There's a lot of cool companies in the capital (not just Shopify!), I'd recommend trying to find a position while you're still a student.

I spoke to a Palantir recruiter once, if you aren't in the Bay Area you'll be doing customisation for customers - not the core engineering stuff that's really cool. Do you know what team at FB you'd be working with? For me the job itself is the most important thing: I wouldn't want to sling PHP, but FB has amazing teams who wrote Hive, Scribe, Presto, HHVM, etc.

CPAN is much higher quality, though. For every Perl package that works and has unit tests and documentation, there are 10 node packages someone tried for a month to write, then gave up and left at 0.01 with no docs. Perl isn't sexy, but Perl diehards have written and published modules to do everything you can imagine.

I didn't think of the box versus letter issue as well. Fetching parcels from my local post office is a 20 minute walk, it can be annoying after a long day of work.

That said, based on the testimony I've seen here I might try a few months and see...

This. I wanted so badly to sign up, my girlfriend and I go to Chinese grocery stores and stock up on junk food a few times a year. Getting top ups and bring exposed to Japanese products would be awesome, but $25 a month is pretty dear for a monthly product. At $10/month for less frequent candy it would be a nobrainer for me.

It'd be nice, but you could fit 40 Canadas into China (population-wise), and most of us already have credit cards due to our proximity with the US. It's sad but completely rational that so many American services don't bother with us as a country.

Another commenter pointed this out, but what you're trying to compute is cosine similarity, in which case you're missing the normalizing part in the denominator (the product of the magnitude of both vectors). In other words, two items which both occur frequently will score higher than two items which occur infrequently, but which co-occur higher than usual. This leads to a tendency to over-recommend popular items.

When you were on EMR, you could have used Mahout's distributed collaborative filtering, which has the benefits of being correct, and requiring zero coding.

Wikipedia explains here: http://en.wikipedia.org/wiki/Cosine_similarity

Much like the Clouderan commenter, I wouldn't put a lot of stock in Berkeley's Big Data Benchmark. I reran a similar test with columnar storage and found Impala handily beats Shark. Operationally it's also much easier to deploy (provided you're on EMR or CDH). The "dedicated nodes" argument is kind of FUD, you can use LLAMA for resource sharing, and you need to colocate imapalad with DataNodes to achieve decent performance anyways. So YARN, Spark and Impala can all play nice together on the same cluster.

The queries and data set Berkeley chooses are bizarre. TPC-DS or TPC-H are much more representative of real-world performance, and the differences are more pronounced as the queries get more complex.

edit: I also don't understand why the Spark team is reinventing the wheel for Spark SQL when Hive running on Tez will produce very similar query plans. The two projects are converging to the same place, but they insist on having divergent code bases ;)