| posted highways speeds are 55 to 65(mph), so traffic is usually in the range of 70 to 85mph, and doesn't really differentiate between the two limits except in the presence of radar. I tend to stick to 60 in the 55 (risk reduction) and 65 in the 65. Lower TCO: no speeding tickets!
Which of course, completely ignores my reality: My job, my kids schools, and my house have a geographic arrangement such that my commute is almost entirely urban. Slower acceleration from a rearward "nose position" with rapid shifting is what has worked best for me: more space so I can time my response better and thus have more time for accelerating than the traffic around me. Then I can still time the lights when in districts where the traffic engineering allows. I also watch far ahead, so I rarely arrive at a red or a queue anywhere near full speed. The "long view" does a lot to eliminate both inefficiency and wear-and-tear.
Most automobiles destined for US markets have had their designs factory tuned to peak efficiency at 60mph, so they look good on the official tests. I expect new models to be a little better at higher speeds, since there is, in fact, now a new segment at 80 mph. But it'll be a while before I drive anything that new!
So... as a side question, can you recommend a good software package or library for clustering moderately short sections of large datasets, with good results for suggesting the dimensionality of the space the resulting clusters should occupy? That would be awesome! |
