https://github.com/tjmahr/fitness
dumping ground for some of my fitness data
https://github.com/tjmahr/fitness
Last synced: 3 months ago
JSON representation
dumping ground for some of my fitness data
- Host: GitHub
- URL: https://github.com/tjmahr/fitness
- Owner: tjmahr
- License: gpl-3.0
- Created: 2022-08-16T18:11:05.000Z (almost 3 years ago)
- Default Branch: main
- Last Pushed: 2023-06-22T19:28:05.000Z (almost 2 years ago)
- Last Synced: 2025-02-24T13:56:36.895Z (3 months ago)
- Language: R
- Size: 27.6 MB
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.Rmd
- License: LICENSE
Awesome Lists containing this project
README
---
output: github_document
---
```{r}
#| include: false
library(tidyverse)
knitr::opts_chunk$set(
collapse = TRUE,
comment = "#>",
dpi = 300,
dev = "ragg_png"
)
# these are fake-ish targets to force a dependence between the data and
# the README
targets::tar_load(data_exercise)
targets::tar_load(data_intervals)
targets::tar_load(data_miles_to_schools)
targets::tar_load(data_bodyweight)
```
# fitness
A dumping ground for some of my fitness data.
## exercise-minutes
I exported all of my Apple Health data. The export process provides a .zip file.
I opened the `export.xml` file in the .zip archive and extracted my daily
exercise minutes (the green ring) activity using the following:
```{r}
#| eval: false
library(tidyverse)
# Get all nodes in the document with summary
nodes_summary <- xml2::read_xml("C:/Users/trist/Desktop/export.xml") |>
xml2::xml_find_all(".//ActivitySummary")
# Get the attributes in each node and create a dataframe from them
nodes_summary |>
xml2::xml_attrs() |>
lapply(as.list) |>
lapply(as.data.frame) |>
bind_rows() |>
as_tibble() |>
# I got my first Apple Watch around Christmas 2018
filter(dateComponents > "2018-12-19") |>
select(
date = dateComponents,
exercise_minutes = appleExerciseTime
) |>
type_convert(
col_types = cols(
date = col_date(format = ""),
exercise_minutes = col_double(),
)
) |>
write_csv("data/exercise-minutes.csv")
```
Now, I have my exercise minutes for each day.
I should disclose that I turn on an Apple workout on my watch when I mow the
lawn, shovel snow, or do something mildly active like going for a walk with my
family.
```{r exercise-minutes, fig.height = 7, fig.width = 10}
library(tidyverse)
data_minutes <- "data/exercise-minutes.csv" |>
# I added entries with Excel and it mangled the dates
read_csv(
col_types = cols(
date = col_date("%m/%d/%Y"),
exercise_minutes = "i")
) |>
filter(!is.na(exercise_minutes)) |>
# in case i want to aggregate by months within years
mutate(
year = lubridate::year(date),
month = lubridate::month(date),
year_month = paste0(year, month)
) |>
group_by(year_month) |>
mutate(
year_month_start = min(date)
) |>
ungroup()
ggplot(data_minutes |> filter(year > 2018)) +
aes(x = date, y = exercise_minutes) +
geom_hline(
yintercept = 30,
color = "#92E82A",
linewidth = 1,
) +
geom_point(
data = function(x) filter(x, exercise_minutes < 125),
alpha = .4
) +
geom_point(
aes(y = 125),
data = function(x) filter(x, exercise_minutes >= 125),
alpha = .4,
color = "orange"
) +
geom_blank(
data = tibble::tibble(
exercise_minutes = NA_real_,
date = as.Date("2023-12-31"),
year = 2023
)
) +
facet_wrap("year", scales = "free_x") +
stat_smooth(method = "loess", formula = y ~ x) +
scale_x_date(date_labels = "%b", date_minor_breaks = "1 months") +
scale_y_continuous(breaks = c(0, 30, 60, 90, 120)) +
labs(
x = "month",
y = "exercise activity minutes ⌚",
caption = "Outliers (125+ minutes) replaced with orange points."
) +
theme_light(base_size = 16)
```
I think that the smoothing lines should be match up from facet to facet. Let's
figure out how to do that.
```{r exercise-minutes-2, fig.height = 7, fig.width = 10}
gam <- mgcv::gam(
exercise_minutes ~ s(as.numeric(date), bs = "cr"),
data = data_minutes
)
data_gam <- gratia::fitted_values(gam, data = data_minutes)
ggplot(data_gam |> filter(year > 2018)) +
aes(x = date, y = exercise_minutes) +
geom_hline(
yintercept = 30,
color = "#92E82A",
linewidth = 1,
) +
geom_point(
data = function(x) filter(x, exercise_minutes < 125),
alpha = .4
) +
geom_point(
aes(y = 125),
data = function(x) filter(x, exercise_minutes >= 125),
alpha = .4,
color = "orange"
) +
geom_blank(
data = tibble::tibble(
exercise_minutes = NA_real_,
date = as.Date("2023-12-31"),
year = 2023
)
) +
geom_smooth(
aes(y = fitted, ymin = lower, ymax = upper),
stat = "identity"
) +
facet_wrap("year", scales = "free_x") +
scale_x_date(date_labels = "%b", date_minor_breaks = "1 months") +
scale_y_continuous(breaks = c(0, 30, 60, 90, 120)) +
labs(
x = "month",
y = "exercise activity minutes ⌚",
caption = "Outliers (125+ minutes) replaced with orange points."
) +
theme_light(base_size = 16)
```
What is my exercise debt (unmet goal minutes)?
```{r}
data_minutes |>
group_by(year) |>
summarise(
n_days = n(),
mean_exercise = mean(exercise_minutes),
exercise = sum(exercise_minutes),
debt = sum(30 - exercise_minutes)
)
data_minutes |>
summarise(
n_days = n(),
mean_exercise = mean(exercise_minutes),
exercise = sum(exercise_minutes),
debt = sum(30 - exercise_minutes)
)
```
My stupid Apple Fitness said my September 2022 challenge is 64 minutes of
exercise activity per day on average, so let's make a query to check on that.
```{r}
data_minutes |>
filter(year == 2022) |>
group_by(year, month) |>
summarise(
num_days = n(),
sum_minutes = sum(exercise_minutes),
mean_minutes = scales::label_number(.1)(mean(exercise_minutes)),
prop_of_sept_22_goal = (sum(exercise_minutes) / (64 * 30)) |> round(2),
.groups = "drop"
) |>
ungroup() |>
arrange(desc(year), desc(month)) |>
knitr::kable(align = "lrrrr")
```
## bodyweight
In April, I took up a new hobby: Weekly boxing classes. This new activity
motivated me to be a little more active (closing my watch's daily exercise goal
more often or trying exercises that help for boxing conditioning). I started
measuring my bodyweight around this time. I haven't change my diet or anything
like that, and I don't have any weight goals besides getting back to my
pre-pandemic bodyweight (250 lbs).
```{r bodyweight, fig.width = 6, fig.height = 4}
data_weight <- "data/bodyweight.csv" |>
read_csv(col_types = cols(date = col_date("%m/%d/%Y")))
p <- data_weight |>
ggplot() +
aes(x = date, y = weight) +
stat_smooth(method = "loess", formula = y ~ x) +
geom_point() +
ylim(225, 270) +
labs(y = "bodyweight [lb]") +
theme_grey(base_size = 16) +
theme(
panel.grid.minor.y = element_blank()
)
p
library(ggrepel)
data_weight |>
ggplot() +
aes(x = date, y = weight) +
geom_hline(
yintercept = 250,
linewidth = 1.5,
color = "grey60"
) +
geom_vline(
linetype = "dashed",
xintercept = as.Date("2022-04-24"),
linewidth = 1,
) +
stat_smooth(method = "loess", formula = y ~ x) +
geom_point() +
ylim(225, 270) +
labs(y = "bodyweight [lb]") +
theme_grey(base_size = 16) +
theme(
# panel.grid.minor.y = element_blank()
) +
geom_text_repel(
data = data.frame(
date = as.Date("2022-04-25"),
weight = 269
),
nudge_y = .5,
point.padding = 0.2,
segment.curvature = 1e-20,
xlim = c(as.Date("2022-04-30"), NA),
label = "Started taking boxing classes in late April 🥊",
size = 4.5
) +
geom_text_repel(
data = data.frame(
date = as.Date("2022-04-26"),
weight = 250
),
nudge_y = 0,
point.padding = 0.5,
segment.curvature = 1e-20,
xlim = c(as.Date("2022-04-30"), NA),
label = "Pre-pandemic weight\naround 250 lbs",
size = 4.5
) +
geom_blank(
aes(x = max(c(data_minutes$date, data_weight$date)), y = 260),
)
# A version with only interesting jumps in weight
last_plot() +
list(data_weight |> filter(abs(weight - lead(weight)) >= 1))
# list()
# last_plot() +
# geom_vline(
# linetype = "dashed",
# xintercept = as.Date("2023-02-16"),
# linewidth = 1,
# ) +
# geom_text_repel(
# data = data.frame(
# date = as.Date("2023-02-20"),
# weight = 257
# ),
# nudge_y = .5,
# point.padding = 0.2,
# segment.curvature = 1e-20,
# xlim = c(as.Date("2023-02-20"), NA),
# label = "Intermittent fasting\non weekdays",
# size = 3.5
# )
```
The last time I was on a months-long fitness kick, I did a Nike+ Run Club
training program for a half-marathon. In a one-month period (July 2015), I ran a
half-marathon distance three times. My weight at that time was 234 lbs, so
that's the benchmark for Tristan, The Cardio Machine.
## other notes
11/25/2022 - got the cross-rope "get lean" weighted jump rope set
12/5/2022 - increased my apple move goal from 860 to 940
01/10/2023 - decreased my apple move goal to 500 because of a family vacation
and procedure. started bumping it back up in 02/2023.
02/20/2023 - started doing intermittent fasting on weekdays. basically, this
kept me from eating peanut butter before bed or in the middle of the night.
### fast miles
In September 2022, I started jogging my kid to preschool in a running stroller.
School is just a little more than a mile away. After a few weeks, I decided to
compete with myself for how fast I could run the first mile split on these jogs.
```{r fast-miles, fig.width = 6, fig.height = 3}
convert_min_sec_to_sec <- function(xs) {
# ye olde splitten applyen combinen
xs |>
strsplit(split = ":", fixed = TRUE) |>
lapply(function(x) {
x <- as.numeric(x)
60 * x[1] + x[2]
}) |>
unlist(use.names = FALSE)
}
data_miles <- "data/miles-to-school.csv" |>
# I added entries with Excel and it mangled the dates
read_csv(
col_types = cols(
date = col_date("%m/%d/%Y"),
first_mile_split = "c",
comment = "c",
season = "n"
)
) |>
mutate(
time = convert_min_sec_to_sec(first_mile_split)
) |>
filter(
str_detect(tolower(comment), "injured", negate = TRUE),
str_detect(tolower(comment), "under the weather", negate = TRUE),
)
ggplot(data_miles) +
aes(x = date, y = time) +
geom_point(size = 3) +
# add some extra spacing in panel as we wait for data
geom_blank(
data = data.frame(
season = 2,
date = as.Date("2023-06-15"),
time = 450
)
) +
scale_x_date(
date_breaks = "2 weeks",
date_labels = "%b %d"
) +
scale_y_continuous(
breaks = (14:22 / 2) * 60,
labels = function(x) {
sprintf("%02d:%02d", x %/% 60, x %% 60)
}
) +
labs(
x = NULL,
y = "pace [min per mi]"
) +
facet_wrap("season", labeller = label_both, scales = "free_x") +
theme_light(base_size = 11) +
ggtitle("🏃♂️ How fast I ran 1 mile, running my kid to school 🎒")
```
## archived/iced things
### intervals
- Run some distance as fast as you can (time it), stop and recover until
heartrate is 140 bpm (time how long the recovery takes).
- Do this for the following distances (m): 800, 800, 600, 600, 400, 400, 200.
I injured my knee doing the 200 m sprint on 2022-09-06 and now I'm kind of
scared of this workout.
```{r intervals, fig.width = 8, fig.height = 4}
library(tidyverse)
d <- "data/intervals.csv" |>
read_csv(
col_types = cols(
interval_number = "i",
time = "c",
distance = "c",
date = col_date("%m/%d/%Y")
)
) |>
mutate(
time = convert_min_sec_to_sec(time)
)
d_date_margins <- d |>
group_by(date, interval_type) |>
summarise(
intervals = n(),
total_time = sum(time),
distance = sum(as.numeric(distance)),
.groups = "drop"
)
d_date_margins
d |>
filter(interval_type == "run") |>
ggplot() +
aes(
x = (interval_number + 1) / 2,
y = (time / 60) / (as.numeric(distance) / 1000)
) +
geom_line(aes(group = date, color = factor(date))) +
geom_point(aes(color = factor(date))) +
geom_label(
aes(
label = paste0(distance, " m"),
y = 3.5
),
hjust = .5,
stat = "unique"
) +
geom_label(
aes(label = label, y = 3.5),
data = data.frame(interval_number = 15, label = "(all)"),
hjust = .5,
) +
geom_point(
aes(color = factor(date)),
data = d_date_margins |>
filter(interval_type == "run") |>
mutate(interval_number = 15) |>
rename(time = total_time),
) +
labs(
x = "running interval number",
y = "pace [min per km] (lower: better)",
color = "date"
)
d |>
filter(interval_type == "run") |>
ggplot() +
aes(
x = (interval_number + 1) / 2,
y = (time) / (as.numeric(distance) / 200)
) +
geom_line(aes(group = date, color = factor(date))) +
geom_point(aes(color = factor(date))) +
geom_label(
aes(
label = paste0(distance, " m"),
y = 30
),
hjust = .5,
stat = "unique"
) +
geom_label(
aes(label = label, y = 30),
data = data.frame(interval_number = 15, label = "(all)"),
hjust = .5,
) +
geom_point(
aes(color = factor(date)),
data = d_date_margins |>
filter(interval_type == "run") |>
mutate(interval_number = 15) |>
rename(time = total_time),
) +
labs(
x = "running interval number",
y = "seconds per 200 m (lower: better)",
color = "date"
)
d |>
filter(interval_type == "recover") |>
ggplot() +
aes(
x = (interval_number + 1) / 2,
y = time
) +
geom_line(aes(group = date, color = factor(date))) +
geom_point(aes(color = factor(date))) +
geom_label(
aes(
label = paste0(distance, " m"),
y = 45
),
hjust = .5,
stat = "unique"
) +
labs(
x = "running interval number",
y = "recovery duration [s] (lower: better)",
color = "date",
caption = "I used a different workout app on 2022-08-16 🤷♀️"
)
```