Failure prediction from production line data/ Statistical process control I'm a research student looking into predicting product defects from on-line time series data. I'm struggling to find exactly which statistics area I should be looking into to create an error prediction model using my data. The response variable is going to be either fail/pass or possibly number of failures in a time period, therefore I was thinking I should look into logistic regression. However the independent variables are time dependent and although there isn't any autocorrelation in the data, errors that do occur tend to occur close together. I've been looking into statistical process control, but haven't really found a journal/book which I've found useful, I need a book/journal which explains how the statistics are calculated and preferably with a real life example. Any help much appreciated.
 A: The literature developed by the "Six Sigma" school of manufacturing quality control is probably the single, best source for applied significance testing of defects.
http://www.isixsigma.com/new-to-six-sigma/getting-started/what-six-sigma/
Any Operations Research textbook will develop the background necessary for this analysis as well as providing resources for deeper dives. Here's a link to one of them ...
http://www.amazon.com/OPERATIONS-RESEARCH-PRINCIPLES-APPLICATIONS-Srinivasan-ebook/dp/B00K7YGMN8/ref=sr_1_2?s=books&ie=UTF8&qid=1447422683&sr=1-2&keywords=principles+of+operations+research&pebp=1447422696233&perid=1AWG9RQ22A9G8G82D1G4
A subset of this class of models has to do with sequential analysis and changepoint detection using methods such as CUSUM. For an advanced treatment of this see Alexander Tartakovsky's Sequential Analysis or the proceedings to this recent Columbia U workshop ... 
http://www.amazon.com/Sequential-Analysis-Hypothesis-Changepoint-Probability/dp/1439838208/ref=sr_1_1?ie=UTF8&qid=1447423227&sr=8-1&keywords=alexander+tartakovsky
https://sites.google.com/site/iwsm2015/home/Program
Once you're ready to build models, then general overviews of predictive modeling would be good to have under your belt. One recent reference is Max Kuhn's Applied Predictive Modeling which has a boatload of R code loaded into a module named Caret ...
http://www.amazon.com/Applied-Predictive-Modeling-Max-Kuhn/dp/1461468485/ref=sr_1_1?ie=UTF8&qid=1447422881&sr=8-1&keywords=kuhn+applied+predictive+modeling
But the bible in the brave, new world of statistical learning remains Hastie and Tibshirani's Elements of Statistical Learning ...
http://www.amazon.com/The-Elements-Statistical-Learning-Prediction/dp/0387848576/ref=pd_sim_14_2?ie=UTF8&dpID=41LeU3HcBdL&dpSrc=sims&preST=_AC_UL160_SR103%2C160_&refRID=04F8SQZ8CVR6CTQJ928D
