Advanced Data Preparation
The truism that 80 percent of the time invested in real-world machine learning projects is spent on data preparation is so widely cited that it is mostly accepted without question. Earlier chapters of this book helped perpetuate the cliché by stating it as a matter of fact without qualification, and although it is certainly a common experience and perception, it is also an oversimplification, as tends to be the case when generalizing from a statistic. In reality, there is no single, uniform experience for data preparation. Yet, it is indeed true that data prep work almost always involves more effort than anticipated.
Rare is the case in which you will be provided a single CSV formatted text file, which can be easily read into R and processed with just a few lines of R code, as was the case in previous chapters. Instead, necessary data elements are often distributed across databases, which must then be gathered, filtered, reformatted, and combined...
