What learning material would you suggest for a CS person / novice statistician / novice mathematician to get into predictive analytics?
$\begingroup$
$\endgroup$
3
-
$\begingroup$ Possible duplicate: stats.stackexchange.com/questions/652/… $\endgroup$– ShaneAug 22, 2010 at 22:43
-
$\begingroup$ I collected some links here: meta.stats.stackexchange.com/questions/6/… $\endgroup$– arsAug 23, 2010 at 4:47
-
1$\begingroup$ This Q has reappeared on the 'active questions' page as it was 'poked' by the 'Community' background process, one of whose tasks is "Randomly poke old unanswered questions every hour so they get some attention". I don't think this question deserves any more attention. It's community wiki, it received only one up-vote, got a reasonable answer on Aug 26 but the user who asked it was last seen on Aug 24, so it seems unlikely that user is ever going to accept an answer. I suggest it should be closed by a moderator. $\endgroup$– onestopOct 26, 2010 at 6:40
Add a comment
|
6 Answers
$\begingroup$
$\endgroup$
4
There's no need to call it Predictive Analytics :) It already has two names: statistics, and data mining.
Beginner Stats Book: Statistics in Plain English
Advanced Stats Book: Multivariate Analysis, by Hair
Data Mining Book: I still haven't found a great one, but Data Mining by Witten is okay.
Don't get too confused by all the details. There are only so many things you can accomplish in general:
- predict a real number (regression)
- predict a whole number (classification)
- modeling (same as the above two, but the model is understandable by humans)
- group similar observations (clustering)
- group similar factors (factor analysis)
- describe a single factor
- describe the relationship between multiple factors (correlation, association, etc)
- determine if a population value is different from another, based on a sample
- design experiments and calculate sample size
good luck!
-
$\begingroup$ No specific thoughts on the Tibshirani, Hastie book, Elements of Statistical Learning? Very comprehensive overview of various methods. I think that would be a great advanced data mining book, anyway. $\endgroup$– AdamOJan 23, 2013 at 6:39
-
$\begingroup$ "Beginner Stats Book: Statistics in Plain English" There are two different books with that same same title by different authors. Is the one referred to here by Harvey J. Brightman or Timothy C. Urda? $\endgroup$– user19959Jan 23, 2013 at 6:53
-
$\begingroup$ Isn't it also Machine Learning? Also, see this question. $\endgroup$ Mar 4, 2013 at 16:32
-
$\begingroup$ Predict a whole number is not classification but rather models like Poisson, NBD, etc. Please be careful with your terminology where it matters. $\endgroup$ Apr 23, 2013 at 12:56
$\begingroup$
$\endgroup$
1
Go to http://www.vaultanalytics.com/books
They have written a book on what predictive models are, when to use what tests/models, and how to create them in Excel. I'm using it every day in my job. I think it's extremely useful.
-
4$\begingroup$ Critical errors in Excel are well described. See for example biostat.mc.vanderbilt.edu/ExcelProblems. It's difficult to see why Excel would be used when there are much better free resources around (e.g., R). $\endgroup$ Jul 7, 2011 at 19:58
$\begingroup$
$\endgroup$
Reading this one now: Predictive Analytics: Microsoft Excel
By Conrad Carlberg
Published Jul 2, 2012 by Que.
- ISBN-10: 0-7897-4941-6 ISBN-13: 978-0-7897-4941-3
I'm not done reading it yet, but so far its a good introduction to the topic for a non-stat person. It starts pretty basic with both stat concepts and Excel functionality and is building from there.
On the Stats front, its going into a pretty healthy discussion of of using moving averages and smoothing to help determine signal/noise in time series.
On the Excel front, its explaining how to build models using the above concepts (rather than just plunking a typical Excel trendline on a chart), and using some of Excels add-on functionality (e.g. Solver and Data Analysis).
$\begingroup$
$\endgroup$
1
I wrote a book on this topic:
"Predictive Analytics: The Power to Predict Who Will Click, Buy, Lie, or Die", by Eric Siegel, Ph.D. (Wiley, February 2013)
More info: http://www.thepredictionbook.com
The Fiscal Times ran an excerpt as an article: http://www.thefiscaltimes.com/Articles/2013/01/21/The-Real-Story-Behind-Obamas-Election-Victory.aspx
And there are other excerpts available throught the book website above.
Let me know if you have any questions about the book!
-
1$\begingroup$ Eric, for some reason your site is very poorly rated with Web of Trust. Apparently for malware? mywot.com/en/scorecard/… $\endgroup$ Mar 4, 2013 at 16:30
$\begingroup$
$\endgroup$
There are quite a few books around. The above are all pretty good. I've also done a book focusing specifically on predictive analytics in retailing and financial services.
Finlay, Steven (2012). Credit Scoring, Response Modeling and Insurance Rating. A Practical Guide to Forecasting Customer Behavior. Basingstoke: Palgrave Macmillan. ISBN 0-230-34776-2.
Its certainly not a "Hard core" mathematics book, but it does give a basic introduction to they key methods such as logistic regression, neural networks etc. In particular it focuses on the entire model development process. Starting with project planning and going through to implementation and monitoring of the model post live.
$\begingroup$
$\endgroup$
1
Further to my previous note - I'd just like to let people know that my new book: Predictive Analytics, Data Mining and Big Data. Myths, Misconceptions and Methods is now out. Available at amazon and all good book shops:
Eric - your book is recommended reading.
-
2$\begingroup$ Please register & merge your 2 accounts (see here). Then you will be able to edit your previous posts, among other advantages. $\endgroup$ Sep 10, 2014 at 20:38