Are there books on practical machine learning? The math is fine in these books, but does not address the practical side: data analysis, pre-processing, on-line pattern recognition, etc.
As someone who has zero calc training nor linear algebra (some discrete mathematics was all I took in University), what are some recommended start point to most quickly be up to speed to digest the resources posted both in the OP and by other commenters in this thread? Just a bit of background about where I am at math-wise: I tried taking Andrew Ng's ML course, and quickly fell behind starting with the second programming assignment (it was implementing a linear regression algo, I believe).
Andrew Ng's Machine Learning course at Coursera, week 1 contains about 1 hour Linear Algebra review: lectures on vectors, matrices, their multiplication, transpose and inverse.
So do you think these lectures are not enough to bring one up to speed in applying these concepts in linear regression?
Of course, a formally educated person has taken a full semester of Linear Algebra, and solved dozens of homeworks of "transpose this", "invert that" etc. so it's difficult to guess how much homework of the boring kind would be needed before one is able to apply these concepts in problem solving.
11 comments
[ 3.0 ms ] story [ 22.1 ms ] threadhttp://web4.cs.ucl.ac.uk/staff/D.Barber/pmwiki/pmwiki.php?n=...
Was delighted to see a notation list as the second page in the book.
PDF download link:
http://web4.cs.ucl.ac.uk/staff/D.Barber/textbook/270212.pdf
[1] Think Stats: http://www.greenteapress.com/thinkstats/thinkstats.pdf
[2] Think Bayes: http://www.greenteapress.com/thinkbayes/thinkbayes.pdf
ET Jaynes: 'Probability Theory: The Logic of Science': http://bayes.wustl.edu/etj/prob/book.pdf
Are great (and free) as a thorough introduction to bayesian reasoning.
So do you think these lectures are not enough to bring one up to speed in applying these concepts in linear regression?
Of course, a formally educated person has taken a full semester of Linear Algebra, and solved dozens of homeworks of "transpose this", "invert that" etc. so it's difficult to guess how much homework of the boring kind would be needed before one is able to apply these concepts in problem solving.
Elementals of Statistical Learning: http://www-stat.stanford.edu/~tibs/ElemStatLearn/index.html