piemonte.Rmd

---
title: "The Piemonte dataset example"
author: "Elias T Krainski"
date: "Started in 2022. Last update: `r format(Sys.Date(), '%a %d %b, %Y')`"
output:
  rmarkdown::html_vignette:
  rmarkdown::pdf_document:
vignette: >
  %\VignetteIndexEntry{The Piemonte dataset example}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
bibliography: INLAspacetime.bib
---

```{r setup, include = FALSE}
knitr::opts_chunk$set(
  echo = TRUE,
  warning = FALSE,
  cache = !TRUE, 
  cache.path = "cache/",
  collapse = TRUE,
  comment = "#>",
  dev = "png",
  dev.args = list(type = "cairo-png"),
  fig.width = 7,
  fig.height = 5
)
```

# Abstract

In this vignette we illustrate how to fit some of the spacetime models in @lindgren2023, 
for the dataset analysed in @cameletti2013.
To perform this we will use the Bayesian paradigm with the**INLA** package, 
using the features provided by the **inlabru** package to facilitate the coding. 

# Introduction 

## The packages and setup

We start loading the required packages and those for doing 
the visualizations, the **ggplot2** and **patchwork** packages.

```{r packages, results = "hide", cache = FALSE}
library(ggplot2)
library(patchwork)
library(INLA)
library(INLAspacetime)
library(inlabru)
```

We will ask it to return the WAIC, DIC and CPO 
```{r ctrc}
ctrc <- list(
  waic = TRUE,
  dic = TRUE,
  cpo = TRUE)
```

## Getting the dataset

We will use the dataset analysed in @cameletti2013, 
that can be downloaded as follows. 
First, we set the filenames
```{r filenames}
u0 <- paste0(
    "http://inla.r-inla-download.org/",
    "r-inla.org/case-studies/Cameletti2012/")
coofl <- "coordinates.csv"
datafl <- "Piemonte_data_byday.csv"
bordersfl <- "Piemonte_borders.csv"
```

Download and read the borders file
```{r downloadporders}
### get the domain borders
if(!file.exists(bordersfl))
    download.file(paste0(u0, bordersfl), bordersfl)
dim(pborders <- read.csv(bordersfl))
```

Download and read the coordinates file
```{r coords}
### get the coordinates
if(!file.exists(coofl))
    download.file(paste0(u0, coofl), coofl)
dim(locs <- read.csv(coofl))
```

Download and read the dataset
```{r downloaddata}
### get the dataset
if(!file.exists(datafl))
    download.file(paste0(u0, datafl), datafl)
dim(pdata <- read.csv(datafl))
```

Inspect the dataset
```{r dinspect}
head(pdata)
```

Prepare the time to be used
```{r time}
range(pdata$Date <- as.Date(pdata$Date, "%d/%m/%y"))
pdata$time <- as.integer(difftime(
    pdata$Date, min(pdata$Date), units = "days")) + 1
```

Standardize the covariates that will be used in the data analysis 
and define a dataset including the needed information 
where the outcome is the log of `PM10`, 
as used in @cameletti2013.
```{r covars}
### prepare the covariates
xnames <- c("A", "WS", "TEMP", "HMIX", "PREC", "EMI")
xmean <- colMeans(pdata[, xnames])
xsd <- sapply(pdata[xnames], sd)

### prepare the data (st loc, scale covariates and log PM10)
dataf <- data.frame(pdata[c("UTMX", "UTMY", "time")],
                    scale(pdata[xnames], xmean, xsd),
                    y = log(pdata$PM10))
str(dataf)
```

# The data model definition 

We consider the following linear mixed model for the outcome
\[ 
\mathbf{y} = \mathbf{W}\mathbf{\beta} + \mathbf{A}\mathbf{u} + \mathbf{e}
\]
where $\beta$ are fixed effects, or regression coefficients including the intercept, 
for the matrix of covariates $\mathbf{W}$,
$\mathbf{u}$ is the spatio-temporal random effect having the matrix
$\mathbf{A}$ the projector matrix from the discretized domain to the data.
The spatio-temporal random effect $\mathbf{u}$ is defined in a continuous 
spacetime domain being discretized considering meshes over time and space. 
The difference from @cameletti2013 is that we now use the models in 
@lindgren2023 for $\mathbf{u}$. 

Define a temporal mesh, with each knot spaced by `h`, 
where `h = 1` means one per day.
```{r tmesh}
nt <- max(pdata$time)
h <- 1
tmesh <- inla.mesh.1d(
  loc = seq(1, nt + h/2, h), 
  degree = 1)
tmesh$n
```

Define a spatial mesh, the same used in @cameletti2013.
```{r smesh}
smesh <- inla.mesh.2d(
    cbind(locs[,2], locs[,3]),
    loc.domain = pborders,
    max.edge = c(50, 300),
    offset = c(10, 140),
    cutoff = 5,
    min.angle = c(26, 21))
smesh$n
```

Visualize the spatial mesh, the border and the locations.
```{r smeshvis}
par(mfrow = c(1,1), mar = c(0,0,1,0))
plot(smesh, asp = 1)
lines(pborders, lwd = 2, col = "green4")
points(locs[, 2:3], pch = 19, col = "blue")
```

We set the prior for the likelihood precision 
considering a PC-prior, @simpson2017pcprior,
through the following probabilistic statements:
P($\sigma_e > U_{\sigma_e}$) = $\alpha_{\sigma_e}$, 
using $U_{\sigma_e}$ = 1 and $\alpha_{\sigma_e} = 0.05$. 

```{r lprec}
lkprec <- list(
    prec = list(prior = "pcprec", param = c(1, 0.05)))
```

With **inlabru** we can define the likelihood model with 
the `like()` function and use it for fitting models with
different linear predictors later.
```{r lhood}
lhood <- like(
  formula = y ~ .,
  family = "gaussian",
  control.family = list(
    hyper = lkprec),
  data = dataf)
```

The linear predictor, the right-rand side of the formula,
can be defined using the same expression for of the both models 
that we are going to fit and is
```{r mcomp}
M <- ~ -1 + Intercept(1) + A + WS + TEMP + HMIX + PREC + EMI +
    field(list(space = cbind(UTMX, UTMY), 
               time = time),
          model = stmodel)
```

# The spacetime models 

The implementation of the spacetime model uses the `cgeneric` 
interface in **INLA**, see its documentation for details. 
Therefore we have a `C` code to mainly build the precision matrix and 
compute the model parameter priors and compiled as static library. 
We have this code included in the **INLAspacetime** package but it is also 
being copied to the **INLA** package and compiled with the same compilers 
in order to avoid possible mismatches. 
In order to use it, we have to define the matrices and vectors needed, 
including the prior parameter definitions.

The class of models in @lindgren2023 have the spatial range, 
temporal range and marginal standard deviation as parameters. 
We consider the PC-prior, as in @fuglstad2015pcmatern, 
for these parameters defined from the probability statements: 
P($r_s<U_{r_s}$)=$\alpha_{r_s}$, 
P($r_t<U_{r_t}$)=$\alpha_{r_t}$ and 
P($\sigma<U_{\sigma}$)=$\alpha_{\sigma}$.
We consider $U_{r_s}=100$, $U_{r_t}=5$ and $U_{\sigma}=2$.
$\alpha_{r_s}=\alpha_{r_t}=\alpha_{\sigma}=0.05$

The selection of one of the models in @lindgren2023 is by chosing the 
$\alpha_t$, $\alpha_s$ and $\alpha_e$ as integer numbers. 
We will start considering the model $\alpha_t=1$, $\alpha_s=0$ and $\alpha_t=2$, 
which is a model with separable spatio-temporal covariance, 
and then we fit some of the other models later.

```{r stmodeldef, echo = FALSE, eval = FALSE}
stmodel <- stModel.define(
    smesh, tmesh, model,
    control.priors = list(
        prs = c(150, 0.05),
        prt = c(10, 0.05),
        psigma = c(5, 0.05)),
    constr = TRUE)
```

## Defining a particular model

We define an object with the needed 
use the function `stModel.define()` 
where the model is selected considering the values for 
$\alpha_t$, $\alpha_s$ and $\alpha_e$ collapsed. 
In order to illustrate how it is done, 
we can set an overall integrate-to-zero constraint, 
which is not need but helps model components identification.
It uses the weights based on the mesh node volumes,
from both the temporal and spatial meshes.
This can be set automatically when defining the model 
by adding `constr = TRUE`.
```{r m102}
model <- "102"
<<stmodeldef>>
```

Initial values for the hyper-parameters help to fit the models in less computing time. 
It is also important to consider in the light that each dataset has its own parameter scale. 
For example, we have to consider that the spatial domain within a box of 
around $203.7$ by $266.5$ kilometers, which we already did when building the 
mesh and setting the prior for or $r_s$. 

We can set initial values for the log of the parameters so that it would 
take less iterations to converge:
```{r ini}
theta.ini <- c(4, 7, 7, 1)
```

```{r fitcode, echo = FALSE, eval = FALSE}
  bru(M,
      lhood,
      options = list(
          control.mode = list(theta = theta.ini, restart = TRUE),
          control.compute = ctrc))
```

The code to fit the model through **inlabru** is
```{r fit102}
fit102 <- 
  <<fitcode>>
```

Summary of the posterior marginal distributions for the fixed effects 
```{r sfixef}
fit102$summary.fixed[, c(1, 2, 3, 5)]
```

For the hyperparameters, we transform the posterior marginal distributions 
for the model hyperparameters from the ones computed in internal scale, 
$\log(1/\sigma^2_e)$, $\log(r_s)$, $\log(r_t)$ and $\log(\sigma)$,
to the user scale parametrization, $\sigma_e$, $r_s$, $r_t$ and $\sigma$, 
respectivelly.
```{r thyper}
post.h <- list(
  sigma_e = inla.tmarginal(function(x) exp(-x/2), 
                           fit102$internal.marginals.hyperpar[[1]]),
  range_s = inla.tmarginal(function(x) exp(x), 
                           fit102$internal.marginals.hyperpar[[2]]),
  range_t = inla.tmarginal(function(x) exp(x), 
                           fit102$internal.marginals.hyperpar[[3]]),
  sigma_u = inla.tmarginal(function(x) exp(x), 
                           fit102$internal.marginals.hyperpar[[4]])
)
```

Then we compute and show the summary of it
```{r shyper}
shyper <- t(sapply(post.h, function(m) 
  unlist(inla.zmarginal(m, silent = TRUE))))
shyper[, c(1, 2, 3, 7)]
```

However, it is better to look at the posterior marginal itself, 
and we will visualize it later.

The model fitted in @cameletti2013 includes two more covariates and 
setup a model for discrete temporal domain where the temporal correlation 
is modeled as a first order autoregression with parameter $\rho$. 
In the fitted model here is defined considering continuous temporal domain 
with the range parameter $r_s$. 
However, the first order autocorrelation could be taken as 
$\rho = \exp(-h\sqrt{8\nu}/r_s)$, where $h$ is the temporal resolution used in 
the temporal mesh and $\nu$ is equal $0.5$ for the fitted model. 
We can compare ou results with Table 3 in @cameletti2013 with

```{r compare1}
c(shyper[c(1, 4, 2), 1], 
  a = exp(-h * sqrt(8 * 0.5) / shyper[3, 1]))
```

# Comparing different models 

We now fit the model $121$ for $u$ as well, 
we use the same code for building the model matrices 
```{r m121}
model <- "121"
<<stmodeldef>>
```
and use the same code for fitting as follows
```{r fit121}
fit121 <- 
  <<fitcode>>
```

We will join these fits into a list object to make it easier working with it 
```{r results}
results <- list("u102" = fit102, "u121" = fit121)
```

The computing time for each model fit
```{r cpu}
sapply(results, function(r) r$cpu.used)
```
and the number of fn-calls during the optimization are
```{r nfn}
sapply(results, function(r) r$misc$nfunc)
```

The posterior mode for each parameter in each model (in internal scale) are
```{r pmode}
sapply(results, function(r) r$mode$theta)
```

We compute the posterior marginal distribution for the hyper-parameters 
in the user-interpretable scale, like we did before for the first model, with
```{r h2pmd}
posts.h2 <- lapply(1:2, function(m) vector("list", 4L))
for(m in 1:2) {
    posts.h2[[m]]$sigma_e =  
      data.frame(
        parameter = "sigma_e", 
        inla.tmarginal(
          function(x) exp(-x/2), 
          results[[m]]$internal.marginals.hyperpar[[1]]))
    for(p in 2:4) {
      posts.h2[[m]][[p]] <-   
      data.frame(
        parameter = c(NA, "range_s", "range_t", "sigma_u")[p], 
        inla.tmarginal(
          function(x) exp(x), 
          results[[m]]$internal.marginals.hyperpar[[p]])
      )
    }
}
```

Join these all to make visualization easier
```{r hpmds2}
posts.df <- rbind(
  data.frame(model = "102", do.call(rbind, posts.h2[[1]])),
  data.frame(model = "121", do.call(rbind, posts.h2[[2]]))
)

ggplot(posts.df) +
  geom_line(aes(x = x, y = y, group = model, color = model)) +
  ylab("Density") + xlab("") + 
  facet_wrap(~parameter, scales = "free")
```

The comparison of the model parameters of $\mathbf{u}$ 
for different models have to be done in light with the 
covariance functions as illustrated in @lindgren2023. 
The fitted $\sigma_e$ by the different models are comparable
and we can see that when considering model $121$ for $\mathbf{u}$, 
its posterior marginal are concentrated in values lower than 
when considering model $102$. 

We can look at the posterior mean of $u$ from both models and 
see that under model '121' there is a wider spread. 
```{r uhist}
par(mfrow = c(1, 1), mar = c(3, 3, 0, 0.0), mgp = c(2, 1, 0))
uu.hist <- lapply(results, function(r)
    hist(r$summary.random$field$mean,
         -60:60/20, plot = FALSE))
ylm <- range(uu.hist[[1]]$count, uu.hist[[2]]$count)
plot(uu.hist[[1]], ylim = ylm,
     col = rgb(1, 0.1, 0.1, 1.0), border = FALSE, 
     xlab = "u", main = "")
plot(uu.hist[[2]], add = TRUE, col = rgb(0.1, 0.1, 1, 0.5), border = FALSE)
legend("topleft", c("separable", "non-separable"), 
       fill = rgb(c(1,0.1), 0.1, c(0.1, 1), c(1, 0.5)), 
       border = 'transparent', bty = "n")
```

We can also check fitting statistics such as DIC, WAIC, 
the negative of the log of the probability ordinates (LPO) 
and its cross-validated version (LCPO), 
summarized as the mean. 

```{r sfits}
t(sapply(results, function(r) {
  c(DIC = mean(r$dic$local.dic, na.rm = TRUE),
    WAIC = mean(r$waic$local.waic, na.rm = TRUE),
    LPO = -mean(log(r$po$po), na.rm = TRUE), 
    LCPO = -mean(log(r$cpo$cpo), na.rm = TRUE))
}))
```

# The automatic group-leave-out cross validation

One may be interested in evaluating the model prediction. 
The leave-one-out strategy was already available in **INLA** 
since several years ago, see @Heldetal2010cpo for details. 
Recently, an automatic group cross validation strategy was implemented, 
see @zhedongH2023gcpo for details. 

```{r g5cv}
g5cv <- lapply(
  results, inla.group.cv, num.level.sets = 5, 
  strategy = "posterior", size.max = 50)
``` 

We can inspect the detected observations that have the posterior linear predictor 
correlated with each one, including itself.
For 100th observation under model "102" we have
```{r group1m102}
g5cv$u102$group[[100]]
```
and for the result under model "121" we have
```{r group1m121}
g5cv$u121$group[[100]]
```
which has intersection but are not the same, for the model setup used.

We can check which are these observations in the dataset
```{r obs1}
dataf[g5cv$u102$group[[100]]$idx, ]
```
and found that most are at the same locations in nearby time. 

We can compute the negative of the mean of the log score so that lower number is better
```{r nlcv}
sapply(g5cv, function(r) -mean(log(r$cv), na.rm = TRUE))
```

# References