Search found 116 matches

This conversion is standard when applying many types of machine learning algorithms like neural nets which cannot deal easily with nominal values.

For example, say you have a feature COLOR, and it ...

You need to use not Weka clustering but some geographic clustering,
see
http://stackoverflow.com/questions/10108368/detecting-geographic-clusters

http://www.rise-group.org/risem/clusterpy/

It is not easy, but technically possible - one can identify many jewish and muslim family names and you can probably buy some spam email list covering UK. But there would be many false positives.
M ...

check GDELT database: Global Data on Events, Location and Tone, which is an amazing tool for data journalists.

http://www.kdnuggets.com/2013/04/gdelt-global-data-on-events-location-tone.html

What you are looking for - automatically show basic visualizations, display significant correlations, and run some basic tests of significance between different variable interactions - you can get som ...

Have you tried Google N-grams - you can download their data from
http://storage.googleapis.com/books/ngrams/books/datasetsv2.html

Does not matter which books you start with -
Elements of Statistical Learning is very good book but more theoretical than others.

I also like Weka 3 book and
Learning from Data - see http://wor ...

Statistics courses are more important for data mining than analysis of algorithms - check also free courses on machine learning from Coursera
and Caltech

There are many free courses on analytics / data mining - see
http://www.kdnuggets.com/2013/02/online-courses-statistics-machine-learning-analytics.html

http://www.kdnuggets.com/2013/04/caltech-fr ...

yes, there is huge work on predicting time series values and lots of textbooks and courses. Neural networks and other special methods will work well.