[tom_b]

FTA (examples of data science):

  targeting health-care customers for hospitals 
  people who can turn social-media clicks and user-posted 
    photos into monetizable binary code is among the biggest 
    challenges facing U.S. industry
  “sentiment analysis,” or finding a way to quantify how
    many tweets are trashing your company or praising it
  determine how customers prioritized paying bills
  “recommendation engines” - those programs that predict
    what you may want to buy next
  advertising

My background is traditional business intelligence, finding actionable data for high level leaders.

A common response on this thread is that little data science is actually occurring in the business world. It would be incredibly useful to me (and I suspect other readers of HN), to hear from other participants on the thread what data science and methods they are using.

I'll kick it off with data analysis examples from my workplace:

  1 - Analysis of patient accruals to various clinical trials.  
    Mainly tracking against goal numbers.  No statistics or ML.
  2 - Analysis of tissue collection opportunities to answer the question:  
    Are we obtaining samples useful to future research opportunities?  No statistics or ML.
  3 - Creating models to accurately predict patient accrual rates for individual studies across various 
    different variables (race, ethnicity, gender, age).
    Simple statistics, probably just a linear regression model.  This is a new effort.
  4 - a long, on-going, and currently unsuccessful attempt to extract useful data from pathology 
    reports (free text descriptions from pathologists examining various collected tissues for 
    both medical treatment and research purposes)

In addition, I know of a few NLP motivated efforts to train classifiers - say, given a set of 500 manually labeled papers, can a classifier be built that would be effective in bucketing an additional 8,000 papers?