Review of Practical Time Series Analysis
Reviewed by Packt Publishing customer· Review provided by
packtpub.com ·
December 6, 2017I think there are some mistakes in the book, despite that I don't regret buying this book on one of the Packt price campaigns.
On page 143 first ARMA(1,1) is recommended by AIC and a few lines later it's ARMA(1,0).
Then a plot is shown with the following text 'The preceding plot shows a good fitting of the ARMA(1,0) model to predict stock closing
prices:'. What makes it a good fit ? With 1000 datapoints it's hard to evaluate the predictions on the plot.
When looking at fewer points is seems as the prediction is just a delay of the close price. I have generated a sub plot and added delayed prices, they are almost identical with the predicted prices so unless I have made a mistake the fitting seems pretty poor.
Running the examples also have some trouble.
The file WDIData.csv can't be found on github
You must create directories as plots/ch1
In chapter 5 loading the best generated neural network model will generate errors as that filename is hardcoded and the generated filenames is not a deterministic process.
Review of Practical Time Series Analysis
Reviewed by Packt Publishing customer· Review provided by
packtpub.com ·
November 6, 2017I already have some experience with Practical Time Series Analysis. I purchased this book to help me stay update with date with new techniques. It does that well.
Review of Practical Time Series Analysis
Reviewed by Packt Publishing customer· Review provided by
packtpub.com ·
January 20, 2018Awesome deal on quite a few quality textbooks and lectures! Well worth the money, contains quality literature that's very helpful and easy to read.
Review of Practical Time Series Analysis
Reviewed by Packt Publishing customer· Review provided by
packtpub.com ·
January 22, 2018Not as in-depth as a primary source, but good for a refresher & the python examples (via jupyter notebook) are a nice framework to experiment with.