Answer--part 2: What's going on in this file?

As you might have figured, I've been traveling, and now that I'm back, I'm trying to catch up.  

I'm not going to give the final answer(s) to the Challenge yet, but I do want to show you another way to look at the data.  

If you open the KML file with a plain text editor, it looks like this: 

As we've already discussed, this is just a KML file.  If you look at the definition of a KML file, you'll see that this is a TRACK piece of data which is just an AltitudeMode (in this case, "clampToGround") followed by a list of timestamps followed by a coord.  Each coord is longitude, latitude, altitude.  (In that order, with the altitude being optional.)  

Since I wanted to really do something interesting with the data, I first opened it in a text editor (again, my choice was TextWrangler), and then pulled out all of the timestamp, long, lat, altitude data.  (notice... this is the opposite of the "lat/long" order that is often used elsewhere).  

Given this, it was pretty straightforward to extract the data and create a data file consisting of four columns, which I then imported into a spreadsheet.  

Based on this, in the spreadsheet I was able to easily compute the additional columns for "Distance in degrees"  "miles traveled" "elapsed time (seconds)" and then ultimately, "speed (mph)." 

 (note:  to compute "distance in degrees" I just used the Pythagorean formula, A^2 + B^2 = C^2 -- that gives me distance in degrees.  See the formula I used in Column F of the spreadsheet.  I looked up that at this latitude, 1 degree of latitude is roughly 55.3 miles.  That's not exactly right, but it's a good enough approximation for this evaluation since all of the location points are more-or-less at the same latitude.)  

I also wrote a couple of spreadsheet functions to extract the day (where 11/22/2014 is day 0, 11/23/2014 is day 1, etc), and the time from the data.  Using those two pieces of information, I could compute the "timevalue," which is the number-of-the-day + fraction of the day as a decimal value.  That sounds complicated, but think of it this way:  noon on the 0th day is 0 + 0.5  (noon is halfway through the day--6PM on the second day is 1 + 0.75, that is, day 1 + 3/4ths of the day, or 1.75).  

Why do that?  I created the "timevalue" so we could easily make a scatter plot of the timevalue (which handily enough goes from 0 - 6) vs. the speed.  

If you look at the second tab (at the bottom of the spreadsheet--the one labeled "Charts of speed and time"), you'll see this: 

See columns A and B?  I just copied those from the main chart above (x-value and speed).  Use copy and then "paste special" in order to just paste the values of the spreadsheet.  

The scatterplot above is interesting, and odd.  You can see big gaps in the data.  Obviously, the phone is turned off during those intervals--that's when I turned my phone off at night.  (Even if it's out of the service area, it will keep recording data.) 

Handily, those correspond pretty nicely to the night times.  

The odd thing is that one spike on day 4.5 (that is, around noon on day 4).  If that's REALLY my speed, then I've moved really fast (close to 3000 mph!!).  Obviously, this bears some closer inspection.  

But zooming in on this data is kind of a pain in Google Spreadsheets, so I went looking for another tool that lets me zoom in more quickly and easily.  

I exported this data to a tab-separated value (TSV) file (just plain text, with each value separated by a tab) and then imported it into a Plot.ly data set.   Here's what the data looks like imported into a Plot.ly scatterplot diagram. Click here to see this data plot live.  

Now the inter-day gaps are really visible, and it's clear that the huge speed spike at day 4.5 has to be spurious.  

And, when you zoom in (which you can do with the live version above), you can see this on Day 0 (that is, timevalues from 0 - 1.0). 

These values look more reasonable.  It's clear that between 0.1 and 0.2 I was driving somewhere, and driving at speeds up to 75 mph.  (That is, on the freeway.)  The phone was turned off between 0.2 and 0.41, and then more freeway driving ensued.  Same story at 0.65 - 0.7.  

Given this kind of view of the data, can you now figure how how the movements (in space) correlated with speeds?  

As another clue, you can look at this plot: 

This is me traveling right around the middle of the day at speeds between 5 and 18 mph from 5.57 until 5.61 (or, to convert that back into regular clock time, that's 1:50PM - 2:35PM).  That's too fast for walking, too fast for running, and too slow for driving in a car.  What could I have been doing during that time?  

More details revealed tomorrow!  We'll talk about the 3,000 mph data spike.  

Search on.  (And now, with spreadsheets and data visualizations!)