CODEWITHSUNDEEP
pythonpython-2.7python-3.xpandasgps
Dan
Dan

Reputation: 1227

Splitting continuous GPS data into separate journeys

I have a large sample of time stamped GPS data for a number of vehicles in text file format. Each vehicles data has a unique ID. I created a Pandas dataframe easily enough and then realized that the each vehicles GPS data is effectively one continuous track for several months.

What I would like to do is isolate individual journeys by splitting the track where the gap in GPS reporting exceeds a certain delta (e.g. 10 minutes). I don't think I can assume that the position does not change between the end of one journey and the beginning of the next (although it /should/ be very close).

uid   ts                     lat      lon
ABC   2017-01-01 00:00:00    0.0000   0.0000
ABC   2017-01-01 00:00:05    0.0000   0.0100
ABC   2017-01-01 00:00:10    0.0000   0.0200
ABC   2017-01-01 00:10:00    0.0100   0.0300 <--- New Journey. 10 min delta 
ABC   2017-01-01 00:10:05    0.0100   0.0400
ABC   2017-01-01 00:10:10    0.0100   0.0500
ABC   2017-01-01 00:10:15    0.0100   0.0600
DEF   2017-01-01 20:00:00    1.0000   1.0000
DEF   2017-01-01 20:00:05    1.0000   1.0100
DEF   2017-01-01 20:00:10    1.0000   1.0200
DEF   2017-01-01 20:20:00    1.0100   1.0300 <--- New Journey. 20 min delta 
DEF   2017-01-01 20:20:05    1.0100   1.0400
DEF   2017-01-01 20:20:10    1.0100   1.0500
DEF   2017-01-01 20:20:15    1.0100   1.0600

Can anyone suggest how I might efficiently go about isolating separate journeys? A solution with Pandas is absolutely not essential.

Upvotes: 2

Views: 392

Answers (1)

DYZ
DYZ

Reputation: 57033

The following splits the dataframe df into a list of dataframes:

delta = pd.to_timedelta(10, unit='m')

breaks = df['ts'].diff() > delta # Feel free to add other conditions!
#0     False
#....
#6     False
#7      True
#8     False
#9     False
#10     True
#11    False
#12    False
#13    False
#Name: ts, dtype: bool

break_locs = df[breaks].index
#Int64Index([7, 10], dtype='int64')

trips = np.array_split(df, break_locs)
#[   uid                  ts   lat   lon
#0  ABC 2017-01-01 00:00:00  0.00  0.00
#1  ABC 2017-01-01 00:00:05  0.00  0.01
#2  ABC 2017-01-01 00:00:10  0.00  0.02
#3  ABC 2017-01-01 00:10:00  0.01  0.03
#4  ABC 2017-01-01 00:10:05  0.01  0.04
#5  ABC 2017-01-01 00:10:10  0.01  0.05
#6  ABC 2017-01-01 00:10:15  0.01  0.06,    uid                  ts  lat   lon
#7  DEF 2017-01-01 20:00:00  1.0  1.00
#8  DEF 2017-01-01 20:00:05  1.0  1.01
#9  DEF 2017-01-01 20:00:10  1.0  1.02,     uid                  ts   lat   lon
#10  DEF 2017-01-01 20:20:00  1.01  1.03
#11  DEF 2017-01-01 20:20:05  1.01  1.04
#12  DEF 2017-01-01 20:20:10  1.01  1.05
#13  DEF 2017-01-01 20:20:15  1.01  1.06]

len(trips)
#3

Upvotes: 5

Related Questions