achompas

21 ms·

151.
by achompas 10y ago
What frameworks are you thinking of, specifically?
152.
by achompas 10y ago
For those downvoting this comment -- it's absolutely true that model performance is discussed at length early in ML courses (usually in the context of the bias-variance tradeoff). My only quibble would be that precision + recall are on
153.
by achompas 10y ago
I think this is closer to reality. Tim is not competing with many folks like him -- he's a knowledgable, experienced, and capable data scientist with significant infrastructure experience, and a net positive to any team he joins. Some
154.
by achompas 10y ago
The important tool: a PhD in applied mathematics that taught me how to think originally and creatively about problems, to abstract and codify them with mathematics, and to approach them numerically and computationally. What is a first app
155.
by achompas 10y ago
Surprised to see this q on HN of all places. He's a co-founder of arguably the most powerful VC in Silicon Valley. https://en.wikipedia.org/wiki/Ben_Horowitz
156.
by achompas 10y ago
Very difficult, at east for research. Top companies with ML research labs (FB, Google, Uber) are basically externalized academic labs. The heads of these labs are simply bringing talent over from their former departments (NYU, Cambridge&#x2
157.
by achompas 10y ago
scikit-learn is a poster child for best practices in machine learning engineering.
158.
by achompas 10y ago
> Is this related to my comment? It was initially related, yeah, but I realized I had uncharitably read your point. I edited my comment, but not enough. Sorry about that. To be fair, this point is often raised in these threads as "w
159.
by achompas 10y ago
Yeah, that's fair!
160.
by achompas 10y ago
That's only part of it. A good data scientist is also good because they know how to answer hard questions. In those situations math isn't "shitty trivia," but instead a tool to be leveraged against those hard questions.
161.
by achompas 10y ago
These types of posts validate my concern about the people entering my field right now. Data science, as a line of work, is distinct from other technical roles in its focus on creating business value using machine learning and statistics .
162.
by achompas 10y ago
Ah, I get you. I could understand that if, say, someone had not read comments from the NYT's public editor. One example: http://publiceditor.blogs.nytimes.com/2015/11/12/as-print-fa...
163.
by achompas 10y ago
I'm baffled as to how my comments can be read as appeals to authority. A brief Google search surfaces clear comments by the NYT itself on its native ad policy. I also think you're suffering from the conflation of marketing and adv
164.
by achompas 10y ago
I think the opposite is true: the NYT is a traditional institution with strong publishing process and serious concern for "separation of church and state," as publishers call the sales/journalism divide. If someone is publish
165.
by achompas 10y ago
We actually agree. I think there's miscommunication going on here. Two facts: 1. NYT does have a native advertising unit. They produce posts like the one I linked earlier -- that article was sponsored by Cole Haan. 2. The Pearl artic
166.
by achompas 10y ago
Yeah agreed. For the record, they're clearly marked as sponsored posts. The URL is even marked as "paidpost.nytimes.com." Here's an example: http://paidpost.nytimes.com/cole-haan/grit-and-grace.html?
167.
by achompas 10y ago
The key with autocorrelated models is to benchmark them against a naive alternative. I really appreciate Jakob's point that the LSTM might simply be modeling one step ahead using the current data point and Gaussian noise. Such a candid
168.
by achompas 10y ago
Are BeatsX actually available in red? They're postponed to February, and only black and white are listed as colors.
169.
by achompas 10y ago
Further, if cable #2 was purchased at an Apple Store, just return it 7 days later. You have a 14-day return window.
170.
by achompas 10y ago
But it's not "no longer." This machine (it's the 2015 rMBP) has not included a dGPU for as long as I can remember.
171.
by achompas 10y ago
About that, yeah. It took me three years to finish my Masters in CS part-time with a 9-5.
172.
by achompas 10y ago
> What is the use of having hundreds of billions in the bank if you're not investing it in growing or creating product lines? Is this an accurate characterization of Apple's strategy over the last 12-24 months? A lot of stuff h
173.
by achompas 10y ago
We use Airflow a lot at iHeartRadio. Most of our ETL processes are written in Aiflow, and the Data Science team is moving its batch models into Airflow too. Happy to answer any questions.
174.
by achompas 10y ago
Who is a Parisian, though?
175.
by achompas 10y ago
> Immigration isn't a fake issue. Take a walk around Miami or San Diego. Or New York. Or London. Or Paris. Or Malmö. I'm the son of Cuban immigrants, born in Miami, living in New York, and have walked Paris many times. I don&#x
176.
by achompas 10y ago
You're being facetious, but: yes, bendgate has caused problems for many actual people in real life. Read more about touch disease: http://motherboard.vice.com/read/iphone-6-plus-touch-disease...
177.
Mapping the World of Music Using Machine Learning: Part 2(tech.iheart.com)
3 points by achompas 10y ago|1 comments
178.
by achompas 10y ago
Oh, also: if you're in NYC I'd be happy to meet over a coffee/beer to swap stories. Feel free to use the contact info in my profile.
179.
by achompas 10y ago
Yeah, I strongly believe companies haven't (or aren't willing to) figure(d) out the IC track problem for data people in the way they've figured it out for engineers. Part of me wonders if it even makes sense for them to fig
180.
by achompas 10y ago
I honestly fell into it by luck. I moved to NYC, studied machine learning in grad school, networked my ass off, and landed an internship. From there I went full time as something of an ML engineer at a company with a strong tech culture, an

More ›