Astsa

+ added US GDP - quarterly adjusted to 1947-1 to 2018-3
+ slight change to `acf1` and `acf2` so user can change `ylim`
+ `Hare` and `Lynx` the 90-year data sets of snowshoe hare and lynx pelts purchased by the Hudson's Bay Company of Canada - note `Lynx` differs from the R data set `lynx`.
+ `gtemp_land` land only - updated global temps to 2017
+ `gtemp_ocean` open ocean only to 2017
+ added data set `salmon`
+ added source of data to `gnp` man page
+ added `plot` option to `acf1` with default `TRUE` and a few additional minor changes

<b> Time Series Analysis and Its Applications With R Examples -- Edition 4 was written under this version.</b>

+ Version 1.8.1 is here on GitHub.

+ These versions are essentially version 1.7.11 but with some minor
changes to satisfy the CRANks. Also, the GitHub version is slightly improved, so call it v1.8.1. The changes are things no one would notice in places where no one looks.


----------------------------
Versions 1.7 - Dec 2016
&starf; Just for historical record, version 1.7 was when CRAN maintainers got CRANky and started asking for arbitrary changes that no one would notice. I'm talking about asking to change one word in the DESCRIPTION file kind of stupid stuff. This is when the GitHub versions started.

1.7.11.

+ added `ARMAtoAR` to give the pi-weights in the invertible representation of an ARMA model ... this is included mainly for pedagogical reasons

+ changed the `max.lag` default in `acf1` and `acf2` so if the series is
seasonal, you'll see at least 4 seasons by default ... I got tired of typing
`acf2(soi, 48)` in class ... now `acf2(soi)` is the same.


1.7.10. in `sarima.for`, added the option to include regressors in the forecast

1.7.9. changed `na.action` to `na.pass` in `acf1`, `acf2`, and `ccf2`... these used to be `na.fail` which is the R stats package default

1.7.8. updated `tsplot` so the time index can be changed

1.7.7. added `tsplot` to give a nice plot of a univariate time series in one easy command ... works like `plot` for a `ts` object.


1.7.6. added `ccf2`, which plots the sample CCF of two series... it operates like `ccf` but the graphic is nicer

1.7.5. added `acf1` giving the sample ACF of a series without the zero lag... it operates like `acf2` but doesn't give the PACF

1.7.4. added data set `hor`, quarterly Hawaiian Occupancy Rate (% of rooms)

1.7.3. some additons to `acf2` allowing a plot title change, and
an option not to produce a graphic (if you only want to use or see the
values in a nice form)

1.7.2. added `plot.all` option to `sarima.for` so that if TRUE, all the data are plotted in the graphic; otherwise,
only the last 100 observations are plotted. The default is `plot.all=FALSE` because it's easier to see
the forecasts if only 100 observations are plotted.

1.7.1. minor changes to `sarima`

+ diagnostic QQplot used to
depend on `MASS` package until it gave warnings on some
simple examples ... now it's done "inhouse"<br/>
+ changed degrees of freedom calculation (wasn't sure
the commands I used to get it were correct... now I'm sure).
+ made `details=FALSE` also shut off the diagnostic plot, so if you run<br/>
`dog <- sarima(cmort, 1,1,1, details=FALSE)` <br/>
then everything (except the graphic) is stored in `dog` and you won't see any output.

+ fixed `x0n` and `P0n` in `Ksmooth0` and `Ksmooth1` (minor fix)

+ add `box()` after `grid()` on some graphics
because the grid was overplotting the box =(

+ added title option to `sarima()` diagnostic plots indicating
model orders (there by default) ... now if you fit a few models,
it's easy to see which diagnostic plot goes with which model...

+ ... also added pointwise 99.9% confidence bounds to the innovations diagnostic QQ-plot based on asymptotic distn of iid normal order statistics ... and some minor tweaks to the Q-stat plot

+ now show ±1 and ±2 root MSPE bounds to `sarima.for()` and used transparent ribbons to display all error bounds because you can't be too pretty

+ added time series `cpg` (annual median cost per GB of storage) for an easy regression with autocorrelated errors exercise

+ added time series `UnempRate`, which can be taken as an update to `unemp` (still there) - the data are monthly US unemployment rate (% unemployed) from 1948 to Nov 2016.

----------------------------

+ added series `globtemp` and `globtempl` ... they are updates to the global temperature deviation series `gtemp` and `gtemp2`, respectively. `globtempℓ` is land only. `gtemp` and `gtemp2` are still in astsa so as not to cause problems - the temperature data sets were reformulated in 2011 so the series do not match up in the overlapping time frame - see the help file for more info

+ added new series `chicken` ... the price of chicken, which is a decent example of trend stationarity

----------------------------

+ needed a minor fix to `acf2`, so while I'm here:

+ minor tweaks to `acf2` and `sarima.for` displays

+ added `xts` dataset `djia` (Dow Jones) so don't need internet connection to use it (or Yahoo now that Verizon owns it - who knows what will happen???)

+ listed p-values in `sarima` `ttable` because t-tables and p-values go together like horses and carriages, and were popular together around the same time.


---------------------------