Skip to content

Commit

Permalink
Merge pull request #638 from cmu-delphi/krivard/vignettes-unlimited
Browse files Browse the repository at this point in the history 
Switch easy vignette examples to unlimited signals
  • Loading branch information
@capnrefsmmat
capnrefsmmat committed Jun 26, 2023
2 parents 0661258 + ec9ed13 commit 525d946
Show file tree
Hide file tree
Showing 4 changed files with 55 additions and 58 deletions.
16 changes: 8 additions & 8 deletions R-packages/covidcast/vignettes/covidcast.Rmd
View file
Expand Up @@ -38,7 +38,7 @@ the United States between 2020-05-01 and 2020-05-07, we can use
library(covidcast)
library(dplyr)
cli <- covidcast_signal(data_source = "fb-survey", signal = "smoothed_cli",
cli <- covidcast_signal(data_source = "fb-survey", signal = "smoothed_wcli",
start_day = "2020-05-01", end_day = "2020-05-07",
geo_type = "county")
knitr::kable(head(cli))
Expand All @@ -64,7 +64,7 @@ request estimates for states instead of counties, we use the `geo_type`
argument:

```{r, message=FALSE}
cli <- covidcast_signal(data_source = "fb-survey", signal = "smoothed_cli",
cli <- covidcast_signal(data_source = "fb-survey", signal = "smoothed_wcli",
start_day = "2020-05-01", end_day = "2020-05-07",
geo_type = "state")
knitr::kable(head(cli))
Expand All @@ -74,7 +74,7 @@ One can also select a specific geographic region by its ID. For example, this is
the FIPS code for Allegheny County, Pennsylvania:

```{r, message=FALSE}
cli <- covidcast_signal(data_source = "fb-survey", signal = "smoothed_cli",
cli <- covidcast_signal(data_source = "fb-survey", signal = "smoothed_wcli",
start_day = "2020-05-01", end_day = "2020-05-07",
geo_type = "county", geo_value = "42003")
knitr::kable(head(cli))
Expand Down Expand Up @@ -219,7 +219,7 @@ First, we can request the data that was available *as of* a specific date, using
the `as_of` argument:

```{r, message = FALSE}
covidcast_signal(data_source = "doctor-visits", signal = "smoothed_cli",
covidcast_signal(data_source = "doctor-visits", signal = "smoothed_adj_cli",
start_day = "2020-05-01", end_day = "2020-05-01",
geo_type = "state", geo_values = "pa", as_of = "2020-05-07")
```
Expand All @@ -228,7 +228,7 @@ This shows that an estimate of about 2.3% was issued on May 7. If we don't
specify `as_of`, we get the most recent estimate available:

```{r, message = FALSE}
covidcast_signal(data_source = "doctor-visits", signal = "smoothed_cli",
covidcast_signal(data_source = "doctor-visits", signal = "smoothed_adj_cli",
start_day = "2020-05-01", end_day = "2020-05-01",
geo_type = "state", geo_values = "pa")
```
Expand All @@ -246,7 +246,7 @@ By using the `issues` argument, we can request all issues in a certain time
period:

```{r, message = FALSE}
covidcast_signal(data_source = "doctor-visits", signal = "smoothed_cli",
covidcast_signal(data_source = "doctor-visits", signal = "smoothed_adj_cli",
start_day = "2020-05-01", end_day = "2020-05-01",
geo_type = "state", geo_values = "pa",
issues = c("2020-05-01", "2020-05-15")) %>%
Expand All @@ -270,7 +270,7 @@ certain lag. For example, requesting a lag of 7 days means to request only
issues 7 days after the corresponding `time_value`:

```{r, message = FALSE}
covidcast_signal(data_source = "doctor-visits", signal = "smoothed_cli",
covidcast_signal(data_source = "doctor-visits", signal = "smoothed_adj_cli",
start_day = "2020-05-01", end_day = "2020-05-07",
geo_type = "state", geo_values = "pa", lag = 7) %>%
knitr::kable()
Expand All @@ -282,7 +282,7 @@ because the query will only include a result for May 3rd if a value were issued
on May 10th (a 7-day lag), but in fact the value was not updated on that day:

```{r, message = FALSE}
covidcast_signal(data_source = "doctor-visits", signal = "smoothed_cli",
covidcast_signal(data_source = "doctor-visits", signal = "smoothed_adj_cli",
start_day = "2020-05-03", end_day = "2020-05-03",
geo_type = "state", geo_values = "pa",
issues = c("2020-05-09", "2020-05-15")) %>%
Expand Down
2 changes: 1 addition & 1 deletion R-packages/covidcast/vignettes/external-data.Rmd
View file
Expand Up @@ -91,7 +91,7 @@ hospitalizations correlate with outpatient doctor visits with deaths during
October 2020, where we use death data as reported by the API.

```{r, message=FALSE}
deaths <- covidcast_signal("indicator-combination", "deaths_incidence_prop",
deaths <- covidcast_signal("jhu-csse", "deaths_7dav_incidence_prop",
start_day = "2020-10-01",
end_day = "2020-10-31",
geo_type = "state")
Expand Down
6 changes: 3 additions & 3 deletions R-packages/covidcast/vignettes/multi-signals.Rmd
View file
Expand Up @@ -23,9 +23,9 @@ library(covidcast)
start_day <- "2020-06-01"
end_day <- "2020-10-01"
signals <- covidcast_signals(data_source = "usa-facts",
signal = c("confirmed_incidence_num",
"deaths_incidence_num"),
signals <- covidcast_signals(data_source = "jhu-csse",
signal = c("confirmed_7dav_incidence_prop",
"deaths_7dav_incidence_prop"),
start_day = start_day, end_day = end_day,
geo_type = "state", geo_values = "tx")
Expand Down
89 changes: 43 additions & 46 deletions R-packages/covidcast/vignettes/plotting-signals.Rmd
View file
Expand Up @@ -14,23 +14,22 @@ structure is designed to be tidy and easily wrangled using your favorite
packages, but the covidcast package also provides some tools for plotting and
mapping signals in an easy way.

For this vignette, we'll use our [combination
signal](https://cmu-delphi.github.io/delphi-epidata/api/covidcast-signals/indicator-combination-inactive.html#statistical-combination-signals)
as an example; the combination indicator is a statistical combination of several
data sources collected by Delphi, and for every county provides a measure of
factors related to COVID activity. We'll also use incident case counts. Fetching
the data is simple:
For this vignette, we'll use our [doctor visits
signal](https://cmu-delphi.github.io/delphi-epidata/api/covidcast-signals/doctor-visits.html)
as an example; it records the percentage of outpatient doctor visits with COVID
symptom codes, as reported by Delphi's health system partners. We'll also use
incident case counts. Fetching the data is simple:

```{r, message=FALSE}
library(covidcast)
comb <- covidcast_signal(data_source = "indicator-combination",
signal = "nmf_day_doc_fbc_fbs_ght",
dv <- covidcast_signal(data_source = "doctor-visits",
signal = "smoothed_adj_cli",
start_day = "2020-07-01", end_day = "2020-07-14")
summary(comb)
summary(dv)
inum <- covidcast_signal(data_source = "usa-facts",
signal = "confirmed_7dav_incidence_num",
inum <- covidcast_signal(data_source = "jhu-csse",
signal = "confirmed_7dav_incidence_prop",
start_day = "2020-07-01", end_day = "2020-07-14")
summary(inum)
```
Expand All @@ -46,7 +45,7 @@ The default `plot` method for `covidcast_signal` objects,
`usmap` package:

```{r}
plot(comb)
plot(dv)
```

The color scheme is automatically chosen to be similar to that used on the
Expand All @@ -56,8 +55,8 @@ One can choose the day and also choose the color scales, transparency level for
mega counties, and title:

```{r}
plot(comb, time_value = "2020-07-04", choro_col = cm.colors(10), alpha = 0.4,
title = "Combination of COVID-19 indicators on 2020-07-04")
plot(dv, time_value = "2020-07-04", choro_col = cm.colors(10), alpha = 0.4,
title = "COVID doctor visits on 2020-07-04")
```

By providing `breaks` and `colors`, we can create custom color scales, for
Expand All @@ -80,7 +79,7 @@ Lastly, we show how we can use custom breaks to (visually) answer the question:
which counties have cumulative case rates of at least 1/100?

```{r, message=FALSE}
cprop <- covidcast_signal(data_source = "usa-facts",
cprop <- covidcast_signal(data_source = "jhu-csse",
signal = "confirmed_cumulative_prop",
start_day = "2020-07-01", end_day = "2020-07-14")
Expand Down Expand Up @@ -117,7 +116,7 @@ We'd like to compare counts per 100,000 against absolute counts, so we fetch the
proportion signal:

```{r, message=FALSE}
iprop <- covidcast_signal(data_source = "usa-facts",
iprop <- covidcast_signal(data_source = "jhu-csse",
signal = "confirmed_7dav_incidence_prop",
start_day = "2020-07-01", end_day = "2020-07-14")
```
Expand All @@ -144,18 +143,18 @@ grid.arrange(p1, p2, nrow = 1)

## Time series plots

Let's fetch the combination indicator and case counts, but for all states rather
than for all counties. This will make the time series plots more manageable.
Let's fetch the doctor visits and case counts, but for all states rather than
for all counties. This will make the time series plots more manageable.

```{r, message=FALSE}
comb_st <- covidcast_signal(data_source = "indicator-combination",
signal = "nmf_day_doc_fbc_fbs_ght",
start_day = "2020-04-15", end_day = "2020-07-01",
geo_type = "state")
inum_st <- covidcast_signal(data_source = "usa-facts",
signal = "confirmed_7dav_incidence_num",
start_day = "2020-04-15", end_day = "2020-07-01",
geo_type = "state")
dv_st <- covidcast_signal(data_source = "doctor-visits",
signal = "smoothed_adj_cli",
start_day = "2020-04-15", end_day = "2020-07-01",
geo_type = "state")
inum_st <- covidcast_signal(data_source = "jhu-csse",
signal = "confirmed_7dav_incidence_prop",
start_day = "2020-04-15", end_day = "2020-07-01",
geo_type = "state")
```

By default, time series plots show all available data, including all
Expand All @@ -166,14 +165,12 @@ states and plot all data for them:
library(dplyr)
states <- c("ca", "pa", "tx", "ny")
plot(comb_st %>% filter(geo_value %in% states), plot_type = "line")
plot(dv_st %>% filter(geo_value %in% states), plot_type = "line")
plot(inum_st %>% filter(geo_value %in% states), plot_type = "line")
```

Notice how in Texas, the combined indicator rose several weeks in advance of
confirmed cases, suggesting the signal could be predictive. Delphi is
investigating these signals for their usefulness in forecasting, as well as
hotspot detection and will publish results when they are available.
Notice how in Texas, the doctor visits indicator rose several weeks in advance
of confirmed cases, suggesting the signal could be predictive.

## Manual plotting

Expand All @@ -186,18 +183,18 @@ For example:
```{r, warning = FALSE}
library(ggplot2)
comb_md <- covidcast_signal(data_source = "indicator-combination",
signal = "nmf_day_doc_fbc_fbs_ght",
start_day = "2020-06-01", end_day = "2020-07-15",
geo_values = name_to_fips("Miami-Dade"))
inum_md <- covidcast_signal(data_source = "usa-facts",
signal = "confirmed_7dav_incidence_num",
start_day = "2020-06-01", end_day = "2020-07-15",
geo_values = name_to_fips("Miami-Dade"))
dv_md <- covidcast_signal(data_source = "doctor-visits",
signal = "smoothed_adj_cli",
start_day = "2020-06-01", end_day = "2020-07-15",
geo_values = name_to_fips("Miami-Dade"))
inum_md <- covidcast_signal(data_source = "jhu-csse",
signal = "confirmed_7dav_incidence_prop",
start_day = "2020-06-01", end_day = "2020-07-15",
geo_values = name_to_fips("Miami-Dade"))
# Compute the ranges of the two signals
range1 <- inum_md %>% select("value") %>% range
range2 <- comb_md %>% select("value") %>% range
range2 <- dv_md %>% select("value") %>% range
# Function to transform from one range to another
trans <- function(x, from_range, to_range) {
Expand All @@ -209,11 +206,11 @@ trans <- function(x, from_range, to_range) {
trans12 <- function(x) trans(x, range1, range2)
trans21 <- function(x) trans(x, range2, range1)
# Transform the combined signal to the incidence range, then stack
# Transform the doctor visits signal to the incidence range, then stack
# these rowwise into one data frame
df <- select(rbind(comb_md %>% mutate_at("value", trans21),
df <- select(rbind(dv_md %>% mutate_at("value", trans21),
inum_md), c("time_value", "value"))
df$signal <- c(rep("Combined indicator", nrow(comb_md)),
df$signal <- c(rep("Doctor visits", nrow(dv_md)),
rep("New COVID-19 cases", nrow(inum_md)))
# Finally, plot both signals
Expand All @@ -222,11 +219,11 @@ ggplot(df, aes(x = time_value, y = value)) +
geom_line(aes(color = signal)) +
scale_y_continuous(
name = "New COVID-19 cases (7-day trailing average)",
sec.axis = sec_axis(trans12, name = "Combination of COVID-19 indicators")
sec.axis = sec_axis(trans12, name = "Doctor visits")
) +
theme(legend.position = "bottom",
legend.title = ggplot2::element_blank())
```

Again, we see that the combined indicator starts rising several days before the
new COVID-19 cases do, an exciting phenomenon that Delphi is studying now.
Again, we see that the doctor visits indicator starts rising several days before
the new COVID-19 cases do.

0 comments on commit 525d946

Please sign in to comment.