When I have some downtime, I extract my Strava activities from the Strava API to practice minimalist visualizations and view trends in my exercise activities. I stock-pile the visuals here to have a single place where I can refer to what visuals I have previously made, so that I can formulate what to practice on in the future. The repo with my code can be found on GitHub, however, I have included some snippets here as well.
Runs
Identifying my running neighborhood
I filtered down to the top 20% of routes I traversed in the last two years, which is how long I have lived in San Mateo. The code reveals the substantial amount of data wrangling that went into transforming a non-spatial dataframe with Google Polylines into spatial routes (lines).
The maps show that my running neighborhood skirts along the boundaries of San Mateo and Burlingame. This is why I call my neighborhood “Bur Mateo.”
# Wrangle activites to routes
decoded_polyline <- googlePolylines::decode(
activities_summary_polyline$map.summary_polyline)
unlist_polyline <- map_df(decoded_polyline, ~as.data.frame(.x), .id="id")
full_polyline_df <- unlist_polyline %>%
left_join(., activities_summary_polyline_id, by = c("id" = "id_p"))
points_to_line <- function(data, long, lat, id_field = NULL, sort_field = NULL) {
# Convert to SpatialPointsDataFrame
coordinates(data) <- c(long, lat)
# If there is a sort field...
if (!is.null(sort_field)) {
if (!is.null(id_field)) {
data <- data[order(data[[id_field]], data[[sort_field]]), ]
} else {
data <- data[order(data[[sort_field]]), ]
}
}
# If there is only one path...
if (is.null(id_field)) {
lines <- SpatialLines(list(Lines(list(Line(data)), "id")))
return(lines)
# Now, if we have multiple lines...
} else if (!is.null(id_field)) {
# Split into a list by ID field
paths <- sp::split(data, data[[id_field]])
sp_lines <- SpatialLines(list(Lines(list(Line(paths[[1]])), "line1")))
# I like for loops, what can I say...
for (p in 2:length(paths)) {
id <- paste0("line", as.character(p))
l <- SpatialLines(list(Lines(list(Line(paths[[p]])), id)))
sp_lines <- spRbind(sp_lines, l)
}
return(sp_lines)
}
}
burmateo <- full_polyline_df %>%
filter(lon >=-122.37 & lon <= -122.0) %>%
filter(lat >= 37.55 & lat <= 37.65)
v_lines <- points_to_line(data = burmateo, ,
long = "lon",
lat = "lat",
id_field = "id")
v_lines_sf <- as(v_lines,
"sf")
st_crs(v_lines_sf) <- 4326
dark_basemap <- cc_location(loc = raster::extent(v_lines_sf),
#zoom = 12,
get_tiles = "https://api.mapbox.com/styles/v1/mapbox/dark-v10/tiles/{zoom}/{x}/{y}",
access_token = token)
lines_bbox <- st_as_sfc(st_bbox(v_lines_sf))
# Draw Map
tm_static_nhood <-
tm_static_mapbox(location = v_lines_sf %>%
st_transform(7131),
style_url = "mapbox://styles/francinestephens/cli17a3cl00v501pohaykfiqe",
username = user) +
tm_shape(v_lines_sf %>%
st_transform(7131)) +
tm_lines(alpha = 0.4,
col = "steelblue") +
tm_layout(frame = FALSE)
Figure 1: Local Running Routes on a basemap
I wanted to plot the runs against parcel data, which are one of my favorite spatial data types to work with to create a minimalist-look. To capture the subset of parcels in my running neighborhood, I created a bounding to clip the county’s parcels down to just the window that features my neighborhood.
# PARCELS
st_crs(parcels)
parcels_reduced <- parcels %>%
st_transform(., crs = 4326) %>%
filter(city == "SAN MATEO" | city == "BURLINGAME") %>%
st_intersection(., lines_bbox)
parcel_version <-
tm_shape(smc) +
tm_fill(col = "white") +
tm_shape(parcels_reduced) +
tm_fill("MAP_COLORS", palette="Greys", alpha = .25) +
tm_shape(v_lines_sf) +
tm_lines(alpha = 0.4,
col = "steelblue") +
tm_layout(frame = FALSE, bg.color = "white"
)
Figure 2: Minimalist view of runs plotted against parcels
Pandemic runs
The images below are limited to my runs at Stanford and in the Texas Hill Country, which were the two locations that I split time in during this period.
# SET COLORS AND THEMES
stanford_cardinal <- "#8C1515"
dark_green <- "#006400"
colors_runs <- scale_color_manual(values = c(stanford_cardinal,
dark_green)
)
theme_by_location <- theme_void(base_family = "Rockwell",
base_size = 13) +
theme(panel.spacing = unit(0, "lines"),
strip.background = element_blank(),
strip.text = element_blank(),
plot.margin = unit(rep(1, 4), "cm"),
legend.position = "none",
plot.title = element_markdown(hjust = 0.5, vjust = 3)
)
runs_by_location <- ggplot(all_runs) +
geom_path(aes(lng, lat, group = id, color = location), size = 0.35, lineend = "round") +
facet_wrap(~location, scales = 'free') +
labs(title = "Francine's
<span style='color:#8C1515'>Stanford</span> & <span style='color:#006400'>Rockwall Ranch, NBTX</span> Runs",
caption = "Runs as of April 14, 2021") +
theme_by_location +
colors_runs
theme_runs <- theme_void(base_family = "Rockwell",
base_size = 20) +
theme(panel.spacing = unit(0, "lines"),
strip.background = element_blank(),
strip.text = element_blank(),
plot.margin = unit(rep(1, 4), "cm"),
legend.position = "none",
plot.title = element_text(hjust = 0.5, vjust = 3)
)
runs_moonrise <- ggplot(all_runs) +
geom_path(aes(lng, lat, group = id, color = group_col), size = 0.35, lineend = "round") +
facet_wrap(~id, scales = 'free') +
labs(title = "Francine's Runs") +
theme_runs +
scale_color_manual(values=wes_palette(n=5, name="Moonrise3"))
All Activities
The following graphs that document the distance and minutes exercised over the last four years show how much working on my dissertation has reduced my ability to run/bike/hike/walk as far or as frequently as before I got into the weeds of the dissertation. Hopefully, the 2023 trends will start to increase at a high clip and become closer to the exercise levels of early 2022 and 2021.
Distance
activities_data_c <- activities_data %>%
mutate(start_date = as_date(start_date),
year = year(start_date),
day_of_year = yday(start_date),
month = month(start_date),
day = wday(start_date, label = TRUE),
week = week(start_date))
activities_data_c %>%
group_by(year) %>%
arrange(start_date) %>%
mutate(cumulative_distance = cumsum(distance)) %>%
ungroup() %>%
filter(year > 2019) %>%
ggplot(aes(x = day_of_year, y = cumulative_distance, color = factor(year))) +
geom_line() +
scale_y_continuous(labels = scales::comma) +
scale_color_brewer(palette = "Set1") +
labs(title = "Cumulative distance per year",
subtitle = "Last 4 years of runs, bikes, hikes, and walks",
x = "Day of Year",
y = "Cumulative Distance (Miles)",
color = "Year",
caption = "Francine's Strava Data") +
theme_classic()
# CALENDAR HEAT MAP OF DISTANCE COVERED EACH DAY
ggplot(activities_data_c %>%
filter(year > 2018), aes(x = week, y = factor(day))) +
geom_tile(aes(fill = distance)) +
scale_fill_continuous(low = "lightgreen",
high = "red") +
facet_wrap(~ year,
scales = "free_x") +
labs(x = "Week",
y = "",
title = "Calendar Heat Map of Distance in Miles",
caption = "Francine's Strava Data") +
theme_classic()
Time
ggplot(activities_data_c %>%
filter(year > 2018) %>%
group_by(year, week, day) %>%
summarize(time = sum(elapsed_time, na.rm = TRUE)) %>%
ungroup() %>%
mutate(minutes = time/60),
aes(x = week, y = factor(day))) +
geom_tile(aes(fill = minutes)) +
scale_fill_continuous(low = "lightgreen",
high = "red") +
facet_wrap(~ year,
scales = "free_x") +
labs(x = "Week",
y = "",
title = "Calendar Heat Map of Exercise Time in Minutes",
caption = "Francine's Strava Data") +
theme_classic()