Blog Detail
Data Sciences Analytics - The Art and Science of Data Mining and Knowledge Discovery
http://dsanalytics.com/dsblog
Data analytics is the analysis of datasets with whatever statistical analysis, machine learning or data mining technique is appropriate. We are consultants based in Sydney, Australia.
I (John Aitchison) have had more than 20 years experience with, and enjoyment of, the probing, exploration, analysis, modelling, interpretation and reporting of datasets : data sets from market research surveys and elsewhere.
Multivariate statistical analysis with a data mining/machine learning flavor (flavour).
Where my work differs from that of other statisticians is, I suppose, that I work mostly with data that others have collected (it could be market research survey data, or financial datasets, or business records, or databases). And I probably do more programming than most ..applications programming such as accessing the databases, delivering the results in some computer based model or application program, and algorithmic programming - I recently completed a Support Vector Machine implementation in Delphi, and am currently working on a visual/spatial hybrid between classification and clustering.
While I do get involved in research design - and I have a viewpoint on this : research design should be driven by a clear vision of what analyses the generated dataset will permit - and have years of hands on experience in data collection (but we don't do that any more), "statistics" is such a vast and expanding area that I think it sensible to state my area of expertise as "data analytics", rather than general purpose statistical consulting.
Recent Posts
Finding your way around R - reprise
an update - more ways to get around R, handy reference cards, some more links...
Most (Published) Research False?
I ponder on some emerging evidence about how often it is that we get things (research/analysis) just plain wrong, and how we can learn what to trust, and the application of those insights to a qualitative/quantitative/analytic world where "the edge" ...
So, Google uses Turkers?
panels of people making judgements, with the objective of improving automated algorithms .. how well does it work?...
Things that go Bump in the Night – a “Negative” (subtractive) alternative approach to Customer Service Modelling and Monitoring
in which I argue that a better "Customer Service" metric might be derived by starting from an assumed level of goodwill, then subtracting the negative experiences. It is hard to be so much better than the competition, but the downside and the mistake...
Predictable Irrationality - implications for research and analytics?
I argue that good research design and modelling can take advantage of, and incorporate, the phenomena of "predictable irrationality", and canvas some of the issues raised in a recent whitepaper and book....
An interesting Forecasting Competition
A new forecasting competition, with the emphasis on "computational intelligence" forecasting methods, and an interesting data set .. cash withdrawals from 111 ATMs...


It is nice blog to help and tips about machine learning or data mining technique visit http://www.bpojobsathome.com
Posted: November 28th, 2008 | More Reviews From bpojob2008 | Report This Comment