Geolife

Geolife Dataset¶

This GPS trajectory dataset was collected in (Microsoft Research Asia) Geolife project by 182 users in a period of over five years (from April 2007 to August 2012). A GPS trajectory of this dataset is represented by sequence of time-stamped points each of which contains the information of altitude, longitude, latitude. The original dataset contains 17,784 trajectories, ~25M Points with a total distance of 1,292,951 kilometers and a total duration of 50,176 hours. These trajectories were recorded by different GPS loggers and GPS phones, and have a variety of sampling rates. 91.5 percent of the trajectories are logged in a dense representation, e.g. every 1~5 seconds or every 5~10 meters per point. The original dataset was filtered and preprocessed, with further details available in the benchmark publication.

Attributes:

latitude: Latitude in decimal degrees.
longitude: Longitude in decimal degrees.
altitude: Altitude in feet (-777 if not valid).
time: Date and time as a string.
trajectory_id: ID of trajectory that Point belongs to.
user_id: ID of user that reported Point.
crs: WGS 84

Run this notebook in Google Colab:

Remember to install the srai library before running the notebook:

%pip install srai[all]

In [1]:

Copied!





import folium
import h3
from IPython.display import display
from srai.datasets import GeolifeDataset
import folium
import h3
from IPython.display import display
from srai.datasets import GeolifeDataset

In [2]:

Copied!

geolife = GeolifeDataset()
geolife = GeolifeDataset()

In [3]:

Copied!

type(geolife.train_gdf), type(geolife.test_gdf)
type(geolife.train_gdf), type(geolife.test_gdf)

Out[3]:

(NoneType, NoneType)

Get data using .load() method -> Default config (Human Mobility Classification)

In [4]:

Copied!

ds = geolife.load()
ds.keys()
ds = geolife.load()
ds.keys()

Out[4]:

dict_keys(['train', 'test'])

In [5]:

Copied!

type(ds["train"]), type(ds["test"])
type(ds["train"]), type(ds["test"])

Out[5]:

(geopandas.geodataframe.GeoDataFrame, geopandas.geodataframe.GeoDataFrame)

In [6]:

Copied!

ds["train"].head()
ds["train"].head()

Out[6]:

	user_id	trajectory_id	avg_altitude_per_hex	timestamp	geometry	h3_sequence_x	h3_sequence_y
0	[000, 000, 000, 000, 000, 000, 000, 000, 000, ...	00011	[492, 492, 490, 490, 491, 491, 491, 195, 92, 6...	[2008-10-28T08:38:26.000000, 2008-10-28T08:38:...	LINESTRING (116.29707 40.01229, 116.29727 40.0...	[8931aa5051bffff, 8931aa505c7ffff, 8931aa5051b...	[8931aa501a7ffff, 8931aa50cd3ffff, 8931aa52a6b...
1	[000, 000, 000, 000, 000, 000, 000, 000, 000, ...	000122	[-628, -628, 127, 133, 147, 165, 168, 187, 204...	[2009-05-30T11:44:45.000000, 2009-05-30T11:44:...	LINESTRING (116.31498 40.0087, 116.31508 40.00...	[8931aa52a53ffff, 8931aa52acbffff, 8931aa52a53...	[8931aa50b4fffff, 8931aa50b43ffff, 8931aa50b53...
2	[000, 000, 000, 000, 000, 000, 000, 000, 000, ...	000123	[79, 79, 94, 102, 98, 166, 138, 163, 112, 119,...	[2009-05-27T16:01:05.000000, 2009-05-27T16:01:...	LINESTRING (116.31975 40.00765, 116.31974 40.0...	[8931aa52a47ffff, 8931aa52a57ffff, 8931aa52a47...	[8931aa4724fffff, 8931aa409b7ffff, 8931aa4724f...
3	[000, 000, 000, 000, 000, 000, 000, 000, 000, ...	000124	[166, 166, 174, 285, 304, 199, 183, 131, 176, ...	[2009-06-20T10:45:27.000000, 2009-06-20T10:45:...	LINESTRING (116.32737 39.99988, 116.32732 39.9...	[8931aa50cd7ffff, 8931aa50cd3ffff, 8931aa501a7...	[8931aa52a0bffff, 8931aa52a57ffff, 8931aa52a47...
4	[000, 000, 000, 000, 000, 000, 000, 000, 000, ...	000126	[118, 118, 141, 58, 32, 111, 87, 11]	[2009-04-23T04:36:07.000000, 2009-04-23T04:36:...	LINESTRING (116.3272 39.99999, 116.32737 39.99...	[8931aa50cd7ffff, 8931aa52a6fffff, 8931aa52a63...	[8931aa52a47ffff, 8931aa52a47ffff]

Creating your own train_test split based on trajectory duration (version TTE) or length version (HMC).

Downloading version all without passing resolution, will return trajectories as linestring geometries.

In [7]:

Copied!

ds = geolife.load(version="all")
ds.keys()
ds = geolife.load(version="all")
ds.keys()

Out[7]:

dict_keys(['train'])

In [8]:

Copied!

ds["train"].head()
ds["train"].head()

Out[8]:

	trajectory_id	user_id	-	altitude	dayNo	datetime	timestamp	geometry
0	000103	[000, 000, 000, 000, 000, 000, 000, 000, 000, ...	[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...	[71, 64, 119, 179, 184, 174, 190, 193, 195, 19...	[39996.1010416667, 39996.101099537, 39996.1011...	[2009-07-02T10:25:30.000000, 2009-07-02T10:25:...	[2009-07-02T10:25:30.000000, 2009-07-02T10:25:...	LINESTRING (116.30824 39.9966, 116.30829 39.99...
1	00011	[000, 000, 000, 000, 000, 000, 000, 000, 000, ...	[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...	[492, 490, 490, 490, 490, 490, 490, 491, 491, ...	[39749.0266898148, 39749.0267476852, 39749.026...	[2008-10-28T08:38:26.000000, 2008-10-28T08:38:...	[2008-10-28T08:38:26.000000, 2008-10-28T08:38:...	LINESTRING (116.29707 40.01229, 116.29727 40.0...
2	000119	[000, 000, 000, 000, 000, 000, 000, 000, 000, ...	[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...	[492, 489, 492, 492, 491, 490, 490, 488, 487, ...	[39978.2868402778, 39978.2869097222, 39978.286...	[2009-06-14T14:53:03.000000, 2009-06-14T14:53:...	[2009-06-14T14:53:03.000000, 2009-06-14T14:53:...	LINESTRING (116.18539 40.12348, 116.18543 40.1...
3	000122	[000, 000, 000, 000, 000, 000, 000, 000, 000, ...	[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...	[-628, -199, -41, 106, 97, 100, 105, 117, 118,...	[39963.1560763889, 39963.1561342593, 39963.156...	[2009-05-30T11:44:45.000000, 2009-05-30T11:44:...	[2009-05-30T11:44:45.000000, 2009-05-30T11:44:...	LINESTRING (116.31498 40.0087, 116.31508 40.00...
4	000123	[000, 000, 000, 000, 000, 000, 000, 000, 000, ...	[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...	[79, 94, 102, 104, 98, 102, 101, 106, 107, 107...	[39960.3340856481, 39960.3341435185, 39960.334...	[2009-05-27T16:01:05.000000, 2009-05-27T16:01:...	[2009-05-27T16:01:05.000000, 2009-05-27T16:01:...	LINESTRING (116.31975 40.00765, 116.31974 40.0...

Passing resolution parameter is neccessary for generation of trajectory in h3 style.

Resolution parameter is required to create h3 sequences from the linestring geometry.

In [9]:

Copied!

ds = geolife.load(version="all", resolution=10)
ds.keys()
ds = geolife.load(version="all", resolution=10)
ds.keys()

Out[9]:

dict_keys(['train'])

In [10]:

Copied!

ds["train"].head()
ds["train"].head()

Out[10]:

	user_id	trajectory_id	h3_sequence	avg_altitude_per_hex	timestamp	geometry
0	[000, 000, 000, 000, 000, 000, 000, 000, 000, ...	000103	[8a31aa501107fff, 8a31aa50111ffff, 8a31aa50102...	[71, 71, 211, 201, 201, 187, 191, 172, 168, 15...	[2009-07-02T10:25:30.000000, 2009-07-02T10:25:...	LINESTRING (116.30824 39.9966, 116.30829 39.99...
1	[000, 000, 000, 000, 000, 000, 000, 000, 000, ...	00011	[8a31aa5051b7fff, 8a31aa505c5ffff, 8a31aa5051b...	[492, 492, 490, 195, 100, 97, 98, 107, 107, 10...	[2008-10-28T08:38:26.000000, 2008-10-28T08:38:...	LINESTRING (116.29707 40.01229, 116.29727 40.0...
2	[000, 000, 000, 000, 000, 000, 000, 000, 000, ...	000119	[8a31aacb386ffff, 8a31aacb1497fff, 8a31aacb14b...	[492, 492, 490, 453, 174, 141, 140, 147, 145, ...	[2009-06-14T14:53:03.000000, 2009-06-14T14:53:...	LINESTRING (116.18539 40.12348, 116.18543 40.1...
3	[000, 000, 000, 000, 000, 000, 000, 000, 000, ...	000122	[8a31aa52a517fff, 8a31aa52a537fff, 8a31aa52a52...	[-628, -628, -199, -41, 106, 136, 140, 146, 14...	[2009-05-30T11:44:45.000000, 2009-05-30T11:44:...	LINESTRING (116.31498 40.0087, 116.31508 40.00...
4	[000, 000, 000, 000, 000, 000, 000, 000, 000, ...	000123	[8a31aa52a467fff, 8a31aa52a46ffff, 8a31aa52a44...	[79, 79, 107, 97, 97, 114, 137, 169, 188, 176,...	[2009-05-27T16:01:05.000000, 2009-05-27T16:01:...	LINESTRING (116.31975 40.00765, 116.31974 40.0...

In [11]:

Copied!

train, test = geolife.train_test_split(
    target_column="trajectory_id", task="TTE", test_size=0.2, n_bins=3
)
train, test = geolife.train_test_split(
    target_column="trajectory_id", task="TTE", test_size=0.2, n_bins=3
)

Created new train_gdf and test_gdf. Train len: 10968, test len: 2743

In [12]:

Copied!

len(train), len(test)
len(train), len(test)

Out[12]:

(10968, 2743)

In [13]:

Copied!

geolife.resolution
geolife.resolution

Out[13]:

In [14]:

Copied!

geolife.test_gdf.head()
geolife.test_gdf.head()

Out[14]:

	user_id	trajectory_id	h3_sequence	avg_altitude_per_hex	timestamp	geometry
0	[000, 000, 000, 000, 000, 000, 000, 000, 000, ...	000103	[8a31aa501107fff, 8a31aa50111ffff, 8a31aa50102...	[71, 71, 211, 201, 201, 187, 191, 172, 168, 15...	[2009-07-02T10:25:30.000000, 2009-07-02T10:25:...	LINESTRING (116.30824 39.9966, 116.30829 39.99...
2	[000, 000, 000, 000, 000, 000, 000, 000, 000, ...	000119	[8a31aacb386ffff, 8a31aacb1497fff, 8a31aacb14b...	[492, 492, 490, 453, 174, 141, 140, 147, 145, ...	[2009-06-14T14:53:03.000000, 2009-06-14T14:53:...	LINESTRING (116.18539 40.12348, 116.18543 40.1...
10	[000, 000, 000, 000, 000, 000, 000, 000, 000, ...	000138	[8a31aa52a467fff, 8a31aa52a477fff, 8a31aa52a40...	[106, 106, 128, 182, 193, 207, 159, 172, 216, ...	[2009-06-23T12:36:35.000000, 2009-06-23T12:36:...	LINESTRING (116.3197 40.0077, 116.3197 40.0077...
16	[000, 000, 000, 000, 000, 000, 000, 000, 000, ...	000169	[8a31aa50cd67fff, 8a31aa50cd47fff, 8a31aa50cd6...	[153, 153, 172, 168, 167, 159, 131, 80, 164, 1...	[2009-05-13T03:55:58.000000, 2009-05-13T03:55:...	LINESTRING (116.32736 39.99976, 116.32737 39.9...
17	[000, 000, 000, 000, 000, 000, 000, 000, 000, ...	0002	[8a31aa52a18ffff, 8a31aa52a0a7fff, 8a31aa52a0a...	[492, 492, 495, 487, 454, 427, 401, 167, 162, ...	[2009-04-12T08:49:05.000000, 2009-04-12T08:49:...	LINESTRING (116.32177 40.01124, 116.32187 40.0...

In [15]:

Copied!





def visualize_h3_trajectories(
    h3_sequences, map_center=(39.98899, 116.32702), zoom_start=12
):
    """
    Visualize H3 sequences on a Folium map.

    Args:
        h3_sequences (List[List[str]]): A list of H3 sequences (trajectories).
        map_center (Tuple[float, float]): Center of the map (lat, lon).
        zoom_start (int): Initial zoom level.
    """
    m = folium.Map(location=map_center, zoom_start=zoom_start, tiles="cartodbpositron")

    colors = ["red", "blue", "green", "purple", "orange", "darkred", "lightblue"]

    for i, sequence in enumerate(h3_sequences):
        color = colors[i % len(colors)]

        for h3_id in sequence:
            boundary = h3.cell_to_boundary(
                h3_id,
            )
            folium.Polygon(
                locations=boundary, color=color, weight=2, fill=True, fill_opacity=0.3
            ).add_to(m)

    return m


h3_sequences = train["h3_sequence"].tolist()
map_ = visualize_h3_trajectories(h3_sequences[10:20])  # visualize first 10 for speed
display(map_)
def visualize_h3_trajectories(
    h3_sequences, map_center=(39.98899, 116.32702), zoom_start=12
):
    """
    Visualize H3 sequences on a Folium map.

    Args:
        h3_sequences (List[List[str]]): A list of H3 sequences (trajectories).
        map_center (Tuple[float, float]): Center of the map (lat, lon).
        zoom_start (int): Initial zoom level.
    """
    m = folium.Map(location=map_center, zoom_start=zoom_start, tiles="cartodbpositron")

    colors = ["red", "blue", "green", "purple", "orange", "darkred", "lightblue"]

    for i, sequence in enumerate(h3_sequences):
        color = colors[i % len(colors)]

        for h3_id in sequence:
            boundary = h3.cell_to_boundary(
                h3_id,
            )
            folium.Polygon(
                locations=boundary, color=color, weight=2, fill=True, fill_opacity=0.3
            ).add_to(m)

    return m


h3_sequences = train["h3_sequence"].tolist()
map_ = visualize_h3_trajectories(h3_sequences[10:20])  # visualize first 10 for speed
display(map_)

Make this Notebook Trusted to load map: File -> Trust Notebook

In [ ]: