![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
Entry tags:
of work and food and data cleaning
I felt like doing something a bit different for today's meal, and thought I'd write it down before I forget what I did.
I started by cutting a small package worth of cherry tomatoes into halves and tossing them into a pot with a bit of watter and some diced carrot. While that was coming to a boil I ground up a single matzo and mixed it with some spices (powdered garlic, onion, cumin, coriander seed, and a dash of curry powder), a couple of tablespoons of chickpea flour and one tablespoon of ground flax seeds. I mixed that with one egg to make a stiff dough, and set it aside while I used the immersion blender on the veg (which had boiled enough (~15 min?) by then to make it possible to blend them). I then added spices to the broth (basil, thyme, oregano, lemon pepper, garlic powder), rolled the dough into little balls (~ 1.5 cm diameter) and put them into the broth, and put the pot back onto the stove. I left it boil for about 10 minutes, while I cleaned up the mess so far and nibbled on a piece of cheese, then turned the burner off (electric hotplate) and let it sit for another 5-10 minutes while I read for a while. It took much longer than just doing a soup with normal wheat-flour egg noodles (because of the need to crush the matzo, and deciding to blend the broth)--nearly 40 minutes elapsed from start to eating, but it was worth it. Very tasty, and there is enough left over for tomorrow, too. (which is good, because tomorrow is Sunday, which means that it isn't possible to go grocery shopping, so I have to eat what is already in the house).
Yesterday during the day I didn't accomplish much, other than heading out to the store where I'd purchased a lamp the day before and trading it in for one that works this time. the rest of the day time was spent goofing off on line and hanging out with friends. However, starting around 21:00 I finally found my motivation and settled into work. The evening's task is one I'd been putting off, but it really needed doing--I went through all of my various data files I've generated and checked them for consistency. The single file that contains all of my microprobe results now actually has the correct information in the field stating if the analysis was bad or good, and which phase it turned out to be, and the total number of bad and good analyses for each phase (one file for each type) now actually matches the total number of analyses I've done--there are no more cases of a single analysis appearing in both the good and the bad files for a phase.
It is amazing how many little data-entry errors can creep in when the data accumulates a bit here, a bit there, over the course of a year. That major cleaning, which I worked on non-stop and without noticing time was elapsing, took until 03:00, and was a necessary first step to today's task.
However, once I finished working last night I didn't actually choose the sensible course of turning off the computer and heading home. Instead I hung out with![[livejournal.com profile]](https://www.dreamwidth.org/img/external/lj-userinfo.gif)
clovis_t on line, caught up on reading Lj/Facebook/blogs, and finally sent out questions for people in that question meme that has been going around. I finally went home and did my yoga for the day at 32:16, followed by reading my 1000. It is a good thing I define "day" with respect to when I sleep, rather than what the clock happens to say, or my number of days in a row daily yoga and 1000 words a day of geologic literature read wouldn't look nearly so good. (I just checked, I've done daily yoga for 2,177 days in a row, and read my 1000 for 220 days in a row.)
Today I slept in till nearly 13:00, but when I arrived at uni actually got straight to work--today's task was sorting out the problem sample. One of the experiments run by a predecessor a decade ago happens to have two different capsules in a single holder. Based upon the way the names are written on the holder we'd assumed that the one on the left was bulk composition "H", and the other "L". However, I'd noticed some time back that a graph showing the talc compositions for this sample show two very distinct clumps, with the old data from a decade ago plotting in position A for the H composition, and position B for the L composition, while my data plotted the opposite. A few weeks back I tried checking the other mineral phases for that sample, to see if the opposite plotting pattern is consistent, and the sample is just miss-labeled. But when I did I was getting a hodge-podge of results, and gave up in disgust. Strangely, now that I've done the major clean up of the data and I'm plotting *only* the true and correct data, It is working much better. To make it easier on myself, I first changed the names of my analysis points so that instead of "H" and "L" in the names I now have "W" and "J", then I plotted for each sample the old data for H in red triangles and the old data for L in red squares, and my new data for "J" in blue triangles, and my new data for "W" in blue squares. Sure enough, all of the triangles plot in one clump, and all of the squares in the other, for every phase present in this sample. Problem solved. Then it was a simple matter of changing each "J" to an "H" and each "W" to an "L" and update all of the files appropriately (changing the codes to show which symbol mathmatica will use to plot these as appropriate). This task took only 3.5 hours, before taking a food break, and now it is only 22:22--the day is young, and I've got time to go through and use mathmatica to do the calculations on the average composition for each experiment so that I can put them onto the poster for the Budapest conference at the end of the month.
I started by cutting a small package worth of cherry tomatoes into halves and tossing them into a pot with a bit of watter and some diced carrot. While that was coming to a boil I ground up a single matzo and mixed it with some spices (powdered garlic, onion, cumin, coriander seed, and a dash of curry powder), a couple of tablespoons of chickpea flour and one tablespoon of ground flax seeds. I mixed that with one egg to make a stiff dough, and set it aside while I used the immersion blender on the veg (which had boiled enough (~15 min?) by then to make it possible to blend them). I then added spices to the broth (basil, thyme, oregano, lemon pepper, garlic powder), rolled the dough into little balls (~ 1.5 cm diameter) and put them into the broth, and put the pot back onto the stove. I left it boil for about 10 minutes, while I cleaned up the mess so far and nibbled on a piece of cheese, then turned the burner off (electric hotplate) and let it sit for another 5-10 minutes while I read for a while. It took much longer than just doing a soup with normal wheat-flour egg noodles (because of the need to crush the matzo, and deciding to blend the broth)--nearly 40 minutes elapsed from start to eating, but it was worth it. Very tasty, and there is enough left over for tomorrow, too. (which is good, because tomorrow is Sunday, which means that it isn't possible to go grocery shopping, so I have to eat what is already in the house).
Yesterday during the day I didn't accomplish much, other than heading out to the store where I'd purchased a lamp the day before and trading it in for one that works this time. the rest of the day time was spent goofing off on line and hanging out with friends. However, starting around 21:00 I finally found my motivation and settled into work. The evening's task is one I'd been putting off, but it really needed doing--I went through all of my various data files I've generated and checked them for consistency. The single file that contains all of my microprobe results now actually has the correct information in the field stating if the analysis was bad or good, and which phase it turned out to be, and the total number of bad and good analyses for each phase (one file for each type) now actually matches the total number of analyses I've done--there are no more cases of a single analysis appearing in both the good and the bad files for a phase.
It is amazing how many little data-entry errors can creep in when the data accumulates a bit here, a bit there, over the course of a year. That major cleaning, which I worked on non-stop and without noticing time was elapsing, took until 03:00, and was a necessary first step to today's task.
However, once I finished working last night I didn't actually choose the sensible course of turning off the computer and heading home. Instead I hung out with
clovis_t on line, caught up on reading Lj/Facebook/blogs, and finally sent out questions for people in that question meme that has been going around. I finally went home and did my yoga for the day at 32:16, followed by reading my 1000. It is a good thing I define "day" with respect to when I sleep, rather than what the clock happens to say, or my number of days in a row daily yoga and 1000 words a day of geologic literature read wouldn't look nearly so good. (I just checked, I've done daily yoga for 2,177 days in a row, and read my 1000 for 220 days in a row.) Today I slept in till nearly 13:00, but when I arrived at uni actually got straight to work--today's task was sorting out the problem sample. One of the experiments run by a predecessor a decade ago happens to have two different capsules in a single holder. Based upon the way the names are written on the holder we'd assumed that the one on the left was bulk composition "H", and the other "L". However, I'd noticed some time back that a graph showing the talc compositions for this sample show two very distinct clumps, with the old data from a decade ago plotting in position A for the H composition, and position B for the L composition, while my data plotted the opposite. A few weeks back I tried checking the other mineral phases for that sample, to see if the opposite plotting pattern is consistent, and the sample is just miss-labeled. But when I did I was getting a hodge-podge of results, and gave up in disgust. Strangely, now that I've done the major clean up of the data and I'm plotting *only* the true and correct data, It is working much better. To make it easier on myself, I first changed the names of my analysis points so that instead of "H" and "L" in the names I now have "W" and "J", then I plotted for each sample the old data for H in red triangles and the old data for L in red squares, and my new data for "J" in blue triangles, and my new data for "W" in blue squares. Sure enough, all of the triangles plot in one clump, and all of the squares in the other, for every phase present in this sample. Problem solved. Then it was a simple matter of changing each "J" to an "H" and each "W" to an "L" and update all of the files appropriately (changing the codes to show which symbol mathmatica will use to plot these as appropriate). This task took only 3.5 hours, before taking a food break, and now it is only 22:22--the day is young, and I've got time to go through and use mathmatica to do the calculations on the average composition for each experiment so that I can put them onto the poster for the Budapest conference at the end of the month.
no subject
no subject
no subject
no subject