Comments on: Poisson vs Gaussian http://hea-www.harvard.edu/AstroStat/slog/2009/poigauss/ Weaving together Astronomy+Statistics+Computer Science+Engineering+Intrumentation, far beyond the growing borders Fri, 01 Jun 2012 18:47:52 +0000 hourly 1 http://wordpress.org/?v=3.4 By: Alex http://hea-www.harvard.edu/AstroStat/slog/2009/poigauss/comment-page-1/#comment-873 Alex Mon, 13 Apr 2009 17:37:48 +0000 http://hea-www.harvard.edu/AstroStat/slog/?p=2166#comment-873 I personally think that Python with numpy/scipy/matplotlib and the ipython frontend is the best lingua franca for such collaborations. If you're familar with Matlab and R, the transition is very easy. Also, you get real object orientation, a price tag of $0, a solid C API for numerically intensive work, and all of Python's libraries for load/cleaning/dealing with the raw data. And, with RPy, you can use R from Python. Oh, and, on top of that, you get happy astronomers :) ; always a good thing. I personally think that Python with numpy/scipy/matplotlib and the ipython frontend is the best lingua franca for such collaborations. If you’re familar with Matlab and R, the transition is very easy. Also, you get real object orientation, a price tag of $0, a solid C API for numerically intensive work, and all of Python’s libraries for load/cleaning/dealing with the raw data. And, with RPy, you can use R from Python.

Oh, and, on top of that, you get happy astronomers :) ; always a good thing.

]]>
By: hlee http://hea-www.harvard.edu/AstroStat/slog/2009/poigauss/comment-page-1/#comment-872 hlee Fri, 10 Apr 2009 15:53:00 +0000 http://hea-www.harvard.edu/AstroStat/slog/?p=2166#comment-872 Upon reading it, I recall my post and the article therein, covering the topic in general: <b><a href="http://hea-www.harvard.edu/AstroStat/slog/2007/interval-estimation-in-exponential-families/" rel="nofollow">Coverage issues in exponential families</a></b>. The authors, Brown, Cai,and DasGupta discussed biases. Alot to be said related to binning, multinomial, approximation or converenge in order to use chi-square type test statistics and to compensate discreteness in Poisson; instead i'd rather put an irrelevant question. <u>How many statisticians would know IDL?</u> (I believe I'm one of not many). For interdisciplinary collaborations, I think setting up a common language or two could accelerate the progress (a bad and living example is myself). Or setting a consistent fashion across communities of writing pseudo codes as a universal language that can be easily transcribed into your choice of language. R is free and widely used among statisticians (believe it or not, statisticians also use various languages and softwares although astronomers may think we only use R) and I believe the counterpart in astronomy is python. Matlab can be a neutral choice since both communities are exposed to this language and very extensively used in computer science and engineering. Shall we talk about this? <b>[Added]</b> I wonder if there's a way, a plug-in maybe, instead of html code, to insert a little survey and to produce subsequent statistics for a review within blog. Thanks~ Upon reading it, I recall my post and the article therein, covering the topic in general: Coverage issues in exponential families. The authors, Brown, Cai,and DasGupta discussed biases.

Alot to be said related to binning, multinomial, approximation or converenge in order to use chi-square type test statistics and to compensate discreteness in Poisson; instead i’d rather put an irrelevant question. How many statisticians would know IDL? (I believe I’m one of not many). For interdisciplinary collaborations, I think setting up a common language or two could accelerate the progress (a bad and living example is myself). Or setting a consistent fashion across communities of writing pseudo codes as a universal language that can be easily transcribed into your choice of language.

R is free and widely used among statisticians (believe it or not, statisticians also use various languages and softwares although astronomers may think we only use R) and I believe the counterpart in astronomy is python. Matlab can be a neutral choice since both communities are exposed to this language and very extensively used in computer science and engineering. Shall we talk about this?

[Added] I wonder if there’s a way, a plug-in maybe, instead of html code, to insert a little survey and to produce subsequent statistics for a review within blog. Thanks~

]]>