“Applied statisticians are often confronted with statistical inference problems dealing with situations in which there appear to be no data, or data of only limited usefulness.”
An example of ‘Sparsity‘ – in which statisticians find themselves having to deal with datasets which are primarily populated with zeros. That’s to say, they’re ‘sparse’. In such cases, where sparsity looms large, what’s a statistician to do? Improbable turns to the work of Professor Stephen W. Looney (Georgia Regents University, US) who presents a seminar entitled: “Much Ado About Almost Nothing: Methods for Dealing With Limited Data” (March 2013).
Those outside the world of professional statistical analysis might guess that having very little data, or perhaps none at all, could be quite a hindrance when it comes to drawing meaningful inferences. But professor Looney comes to a conclusion which may offer some encouragement for statisticians facing sparsity :
“Even if no data or extremely limited data are present, valid statistical methods are available.”
Also see, related: Non-ignor ble mis ingn ss [note: since publication, some of the page's links have gone mis ing]