automate de-noise process

[swangcs]

To automate the whole data-preprocessing:

• input: raw GPS datasets
• output: multiple bus trips with cleaned trajectory points sequence

I will also be involved when I am back in Dublin.

[Ruixinhua]

I set following rules to filter some useless points:

• filter 1: stop id filter, if stop id that the bus hold is not in the line 15, then skip.
• filter 2: distance filter, the distance which is between bus and bus stop, if the distance is too far(>1000 meters), then skip.
• filter 3: delta time filter, if the delta time between the current bus timestamp and last bus timestamp is too high(>1000 seconds), then skip.
• filter 4: speed filter, if the speed of bus is too high(>120km/h) or too low(<1km/h), then skip.
• filter 5: accumulated distance filter, if the accumulated distance of bus is longer than the total distance of standard trip, then skip.
• filter 6: if the final trajectory is too short(has less than 100 points), then remove the trajectory.
• filter 7: only choose the route begin with a standard start stop, skip the trajectory that is not begin with a start stop. @swangcs see commit c77e07f