not getting sub date granularity in google docs trend lines?

Question

Anyone have an idea what I might have broken that prevents me from getting sub date granularity on google doc's spreadsheet's trend line graphs?

My data is in variable increments, anywhere from 6 to 8 hours apart like so:

1-1-2011 00:00:00     100  
1-2-2011 00:00:00     103  
1-3-2011 00:00:00     107  
1-4-2011 00:00:00     109  
1-4-2011 06:00:00     111  
1-4-2011 10:00:00     120  
1-4-2011 14:00:00     132  
1-4-2011 20:00:00     176  
1-5-2011 00:00:00     190  
1-5-2011 09:30:00     221  
1-5-2011 15:00:00     250  
1-5-2011 19:15:00     272  
etc...

When I build a trend line graph, everyone of those entires on the 4th and 5th are stacked up in a horizontal line, then it jumps over to the 5th where they stack up again. End result is a stair step instead of a nice growth curve.

Here's a screen shot of the resulting graph with my real data (above is a sanitized summary), each green arrow points to a datapoint.

I built a test spreadsheet with the above test data... it of course works like a charm. Even after adding a second series.

Answer 1

Maybe you're using the wrong chart type for display this kind of information. When I selected time line from the Charts for the above data, I'm getting an interactive chart, exactly like the one they are using in Google Finance.

Answer 2

Ok, it appears to be caused by entering just enough data to tip the balance somehow. Throw this data into a sheet and apply a timeline chart. You'll get a nice curved line.

1/11/2011 00:00:00 1173
1/12/2011 00:00:00 1270
1/13/2011 00:00:00 1387
1/14/2011 00:00:00 1434
1/15/2011 00:00:00 1458
1/17/2011 00:00:00 1531
1/18/2011 00:00:00 1567
1/19/2011 00:00:00 1598
1/20/2011 00:00:00 1624
1/21/2011 12:14:00 1643
1/21/2011 12:30:00 1722
1/21/2011 13:30:00 1863
1/21/2011 14:00:00 1895
1/21/2011 16:45:00 1990
1/21/2011 19:23:00 2026
1/21/2011 23:59:59 2060
1/22/2011 11:00:00 2134
1/22/2011 19:46:00 2184
1/23/2011 14:00:00 2232

Now add this additional row:

1/10/2011 00:00:00 893

Boom, stair steps.

(this is a screen shot of the test chart I linked to in the question.)

What's sad is that it appears to be an intentional data simplification. If the data in the cart bounces up and down within a day, then they don't seem to compress it. it's only if there is only growth or decline, then they compress it to day scoped as soon as you get enough data. It does not appear to be related to the total span of the data, as I did experiments with having the 1/10 data in place, and removing the 1/15 data returns you to a smooth curve.

Short of getting someone from Google to comment on this, I don't think we're going to get a better answer. :(