Sacha Chua on Nostr: I got curious and did a scholar.google.com search for "Sacha Chua." The most amusing ...
I got curious and did a scholar.google.com search for "Sacha Chua." The most amusing thing I found was:
Joscha Cueppers and Jilles Vreeken. Just Wait For It... Mining Sequential Patterns with Reliable Prediction Delays. In: IEEE International Conference on Data Mining (ICDM). 2020
which (among other things) had apparently analyzed the time tracking records that I'd intentionally made public:
"Next we consider Lifelog, which is based on the life of
Sacha Chua who logs and publishes all her daily activities. We considered the data over 2017, removing any activities
with have the same start and stop timestamp. As this dataset
provides many events that are potentially interesting, we
consider every e ∈ Ω as target, and have 40 target sequences
with Y [i] = 1 iff X[i] = e. In addition, we consider a Y
where we marked all business related activities as interesting.
Over all these datasets, SCIS discovers on average 695
patterns, many of which are redundant and not all make
intuitive sense. While SQS only discovers 3 predictive patterns, these do make sense: Cook, Dinner→Clean the Kitchen
and Subway, Social→Subway. OMEN takes between 6.1 and
37 seconds per dataset, and overall discovers 24 patterns.
Many of these, such as Sleep→Childcare, Cook→Dinner,
Dinner→Clean the Kitchen, predict the next action, i.e. a time
delay distribution with a peak at 1. A more interesting pattern
is Subway→Subway which has its peak at δ = 2, and for
which a natural interpretation is that Sacha takes the subway,
logs on average one activity, and then takes the subway back."
https://publications.cispa.de/articles/conference_contribution/Just_Wait_For_It_Mining_Sequential_Patterns_with_Reliable_Prediction_Delays/24613377?file=43247712
2017! Bwahaha... I had a one-year-old child and was trying to stay sane by squeezing in some consulting here and there while dealing with sleep deprivation and all sorts of other new-parent challenges. :) I still don't have the time to do lots of different things on one errand, although it's nice that I'm now biking around a lot more than I use the subway. Maybe analyzing 2012-2016 might have been more interesting for their data mining, since that covered a little bit of corporate work time and the transition to self-directed learning in my semi-retirement experiment.
Cool, cool, very fun, I'm tickled pink that someone else found the data nifty. I recently made quantifiedawesome.com more private, but maybe I should open that part up again.
Joscha Cueppers and Jilles Vreeken. Just Wait For It... Mining Sequential Patterns with Reliable Prediction Delays. In: IEEE International Conference on Data Mining (ICDM). 2020
which (among other things) had apparently analyzed the time tracking records that I'd intentionally made public:
"Next we consider Lifelog, which is based on the life of
Sacha Chua who logs and publishes all her daily activities. We considered the data over 2017, removing any activities
with have the same start and stop timestamp. As this dataset
provides many events that are potentially interesting, we
consider every e ∈ Ω as target, and have 40 target sequences
with Y [i] = 1 iff X[i] = e. In addition, we consider a Y
where we marked all business related activities as interesting.
Over all these datasets, SCIS discovers on average 695
patterns, many of which are redundant and not all make
intuitive sense. While SQS only discovers 3 predictive patterns, these do make sense: Cook, Dinner→Clean the Kitchen
and Subway, Social→Subway. OMEN takes between 6.1 and
37 seconds per dataset, and overall discovers 24 patterns.
Many of these, such as Sleep→Childcare, Cook→Dinner,
Dinner→Clean the Kitchen, predict the next action, i.e. a time
delay distribution with a peak at 1. A more interesting pattern
is Subway→Subway which has its peak at δ = 2, and for
which a natural interpretation is that Sacha takes the subway,
logs on average one activity, and then takes the subway back."
https://publications.cispa.de/articles/conference_contribution/Just_Wait_For_It_Mining_Sequential_Patterns_with_Reliable_Prediction_Delays/24613377?file=43247712
2017! Bwahaha... I had a one-year-old child and was trying to stay sane by squeezing in some consulting here and there while dealing with sleep deprivation and all sorts of other new-parent challenges. :) I still don't have the time to do lots of different things on one errand, although it's nice that I'm now biking around a lot more than I use the subway. Maybe analyzing 2012-2016 might have been more interesting for their data mining, since that covered a little bit of corporate work time and the transition to self-directed learning in my semi-retirement experiment.
Cool, cool, very fun, I'm tickled pink that someone else found the data nifty. I recently made quantifiedawesome.com more private, but maybe I should open that part up again.