Peel Brain

org-mode has a feature called captured column view. This allows me to bring the column view table into another org file.

In a file named sleep.org I have this (it's the last header in the file, because this is not the first thing I want to see when I open this file, and it can get very long as the years go on).

* Data
#+BEGIN: columnview :id "file:diary.org" :maxlevel 3 :skip-empty-rows t :match "Move>0"
#+NAME: raw-data
#+END:

To break down the important bits:

:id

this is a reference to the file we are using as a column view. It will use the column view defined at the top of that file, to lay out the columns here, so they don't need to be specified again.

:maxlevel:

This says only go as far as the *** level. That is where all the data we want is, so if I add a fourth level it will not bring it in here.

:match

This is a filter. The column view will bring in levels one and two as well, so by filtering on only records with a move value I will only have the rows with actual data. The decision to filter on move is arbitrary, it could be anything.

Pressing C-c C-c on the #+BEGIN: line will load the table from diary.org. This must be done manually, unlike code blocks which update automatically.

I have a code block that "cleans" the data.
This does a few things, and will be expanded as I want to plot more data points:

• Extracts the year from the date.
• Extracts the location from the tags.
• Converts hours from HH:mm to a decimal.

I am using R ¹ for this. \\

What I like about R is that it references table columns by name. That means, when I add another column (e.g., when I added Year), I don't need to update the column indexes on all the other code blocks.

#+NAME: clean-data
#+begin_src R :var data=raw-data :results value table :colnames yes :exports none
library(tidyverse)
decTime <- function(x) {
  hour <- as.numeric(str_extract(x, "\\d+"))
  min <- as.numeric(str_extract(x, "\\d+$"))
  total <- hour + (min / 60)
  round(total, 2)
}

location <- function(x) {
  if (length(x) > 1) {
    return(sapply(x, location))
  }

  locations <- c("@sochi", "@alberton", "@clarens", "@moscow", "@kruger", "@nizhny", "@air")
  names <- c("Sochi", "Alberton", "Clarens", "Moscow", "Kruger", "Nizhny", "Air")
  idx <- which(sapply(locations, function(pattern) grepl(pattern, x)))
  if (length(idx) > 0) names[idx] else "Unknown"
}

data %>%
  mutate(Hours = decTime(Sleep)) %>%
  mutate(Year = substr(Date, 2, 5)) %>%
  mutate(Location = location(TAGS)) %>%
  filter(Hours > 0) %>%
  select(Year, Date, Location, Caff, Hours)
#+end_src

The name above this block is important, because it allows the results of this block to be used as a table in another block.

I have a few more R code blocks that export charts. Here are a few of them.

Histogram

This is a quick visualization of how many nights I slept for various hour ranges.

#+begin_src R :var data=clean-data :results output graphics file :file hours_histogram.png :width 996 :height 500 :exports results
library(ggplot2)
ggplot(data, aes(x = data$Hours )) +
  geom_histogram(binwidth = 1, fill = "skyblue", color = "black", alpha = 0.7) +
  scale_x_continuous(breaks = 0:10) +
  scale_y_log10() +
  labs(title = "Distribution of Hours",
       x = "Hours Slept",
       y = "") +
  theme_minimal()
#+end_src

Scatter plot

I want to see if caffeine has an impact on my sleep, so I will use a scatter plot.

#+begin_src R :var data=clean-data :results output graphics file :file caffeine_sleep_scatter.png :width 600 :height 400 :exports results
library(ggplot2)

# Calculate correlation coefficient
correlation <- cor(data$Caff, data$Hours, use = "complete.obs")

# Create scatter plot with trend line (Caffeine on X, Sleep on Y)
ggplot(data, aes(x = Caff, y = Hours)) +
  geom_point(alpha = 0.6, color = "steelblue", size = 2) +
  geom_smooth(method = "lm", se = TRUE, color = "red", alpha = 0.3) +
  labs(title = paste("Caffeine vs Sleep Hours (Correlation:", round(correlation, 2), ")"),
       x = "Caffeine (mg)",
       y = "Sleep Hours") +
  theme_minimal() +
  theme(plot.title = element_text(hjust = 0.5))
#+end_src

Box plot

I want to see if I sleep better or worse in different locations. A box plot can show me not only the average sleep by location, but also the range.

#+begin_src R :var data=clean-data :results output graphics file :file hours_boxplot.png :width 996 :height 400 :exports results
library(ggplot2)
ggplot(data, aes(x = Location, y = Hours)) +
  geom_boxplot(fill = "skyblue", alpha = 0.7) +
  labs(title = "Hours Distribution by Location",
       x = "Location",
       y = "Hours") +
  theme_minimal() +
  theme(axis.text.x = element_text(angle = 45, hjust = 1))
#+end_src