Prev: Create an HTML report from this dataset
Next: Passing a non-temporary array to a proc fcmp function
From: Eli Y. Kling on 7 Jan 2010 04:55 Hmmm, MAKING it talk has the connotations of a Water-Boarding. When you torture the data you get Over-Fitting it tells you what you want to hear. I prefer to sit down with my data to a nice dinner and maybe invite it up later for Coffee. Advanced techniques work better than crude force and allow to coax and tease out those shy correlation, interactions and trends. Nerds Rule. On 6 Jan, 21:54, r...(a)CDC.GOV ("Fehd, Ronald J. (CDC/CCHIS/NCPHI)") wrote: > > Making Data Talk > > Communicating Public Health Data to the Public, Policy Makers, and the > Press > > a book by NCI authors Drs. David Nelson, Brad Hesse, and Bob Croyle. > > http://www.amazon.com/Making-Data-Talk-Communicating-Public/dp/019538153 > X > > > > > Presenters will provide practical suggestions > > on how scientists and other public health practitioners > > can better communicate data to the public, policy makers, > > and the press in typical real-world situations.- Hide quoted text - > > - Show quoted text -
From: Frank DiIorio on 7 Jan 2010 06:08 On Jan 7, 4:55 am, "Eli Y. Kling" <eli.kl...(a)gmail.com> wrote: > Hmmm, > > MAKING it talk has the connotations of a Water-Boarding. When you > torture the data you get Over-Fitting it tells you what you want to > hear. I prefer to sit down with my data to a nice dinner and maybe > invite it up later for Coffee. Advanced techniques work better than > crude force and allow to coax and tease out those shy correlation, > interactions and trends. > > Nerds Rule. Maybe so, but I prefer the "24" approach, with Jack Bauer holding a hot iron close to the face of the dataset, screaming "TELL ME NOW! WHERE IS THE CORRELATION?! AND WHY THE F*** ARE MY TABLES SPARSE? WHERE DID YOU PUT THE MISSING VALUES?" Dinner and afterglow are fine, but when you need results in a hurry, I'd go for the Bauer method. I'd also respectfully add that the "coaxing and teasing" methods are manipulative, and are not unlike getting a drunken prom queen to do your bidding. Nerds do, indeed, rule (or maybe we watch too much TV)
|
Pages: 1 Prev: Create an HTML report from this dataset Next: Passing a non-temporary array to a proc fcmp function |