在熊猫数据框中查找模式

Question

I have a sizeable Pandas data frame which looks like this.

                                           id  rssi location         day      time
0        2a463296-bd84-512a-8484-9d79649922ae    58     G-19  2016-01-27  12:35:23
1        c6a18d27-63ba-5457-99c1-4c08f6410e33    74     G-19  2016-01-27  12:35:26
2        ee75fa2d-66d9-52e5-9198-a886288ba044    74     G-19  2016-01-27  12:35:28
3        3dc1f5f5-eab3-541c-97f8-e57f32bdf53d    82     G-19  2016-01-27  12:35:28
4        6c1b9019-a6bc-5ed6-82e6-879b7c120991    62     G-19  2016-01-27  12:35:33
26       2a463296-bd84-512a-8484-9d79649922ae    38     G-20  2016-01-27  12:36:58
27       c6a18d27-63ba-5457-99c1-4c08f6410e33    70     G-20  2016-01-27  12:36:59
28       7edb5047-62b8-58bf-89f4-4151d7b694f4    70     G-20  2016-01-27  12:37:01
29       f4c906a8-7680-5bac-b7a0-be408364a268    58     G-20  2016-01-27  12:37:07

... ... ... ... ... ...

1546516  c6a18d27-63ba-5457-99c1-4c08f6410e33    58     G-59  2016-01-27  13:53:44
1546517  2a463296-bd84-512a-8484-9d79649922ae    50     G-59  2016-01-27  13:53:48
1546518  10baa504-7eec-522f-990b-61b3c215352d    50     G-59  2016-01-27  13:53:49
1546519  15ce7c62-3014-5734-9025-b658278cd33a    42     G-59  2016-01-27  13:53:51
1546520  54b281f5-e532-5fd8-b681-e5bffcd4d6bb    62     G-59  2016-01-27  13:53:53
1546521  1300368f-c823-5fa7-8241-0b245f601859    46     G-59  2016-01-27  13:53:55
1546522  79f64138-d332-51c8-a583-686f30eb65f9    70     G-59  2016-01-27  13:53:56

Each id is the id of a WiFi device. I am trying to build up picture of the movements of the device. For example I want to model how 2a463296-bd84-512a-8484-9d79649922ae went from G-19 to to G59 but spent over 1 hour in G20 so we presume the device just passed G19.

From tests I carried out most devices will be seen every 5 minutes at a minimum. Obviously as a device passes one of the detectors it may not be seen as its only detected when it broadcasts a beacon for WiFi.

I want to be able to show that device A was in location 1 at for x amount of time and then went to location 2 for y amount of time or passed location c by the exit door and wasn't seen again for a number of hours.

I am only worried about what a device is doing on a particular day and not any other day.

What's the best way of going about solving this.

Answer 1

Use boolean selection to grab the item you want by ID. Select further using the date as criteria. Do math using the time column. If you want to see when the device wasn't visible at all, it should be apparent when you look at a specific device after selection by ID.

Please see pandas docs for details. pandas Indexing and Selecting

Edit:

for i in df["day"].unique():
    date_df = df[df["day"]== i]
    ids_seen = set()
    for index, row in date_df.iterrows():
        if row["id"] not in ids_seen:
        '''
          Enter nested loop and do some stuff for each unique id. Add it to a set so we can keep track
        '''
        ids_seen.add(row["id"])