WEBVTT 00:01.320 --> 00:08.050 In this video we learn how to index lost life time series for particular timestamps or time periods. 00:08.100 --> 00:14.960 So for example we can only select the temperance information for the year 2015 or for the month made 00:14.980 --> 00:16.370 to 14. 00:16.380 --> 00:17.930 And we can simply do this. 00:17.920 --> 00:19.860 This is the loke operator. 00:20.340 --> 00:27.930 But first of all let's import pandas and for indexing and slicing time series we need to have a date 00:27.930 --> 00:30.270 time indexed in our data frame. 00:30.350 --> 00:32.330 And this is exactly what we are doing here. 00:32.360 --> 00:38.590 So we are importing our CSP fire temperature and we are passing the column daytime to the parameters 00:38.590 --> 00:43.110 past states and index columns. 00:43.310 --> 00:49.470 So now give we f on the left hand side our daytime index and we can also confirm this with the input 00:49.490 --> 00:50.380 method. 00:50.390 --> 00:52.900 So here we have the daytime index. 00:53.330 --> 01:00.770 And same with all other indexes we can select and slice for index labels by using the local operator 01:00.970 --> 01:02.500 service for the log operator. 01:02.530 --> 01:09.290 We can perform labor based indexing and give us a daytime and next hour labors are actually here day 01:09.290 --> 01:10.270 times. 01:10.430 --> 01:16.070 And let's assume that we only want to select the one time stamps or flag sample here the first of January 01:16.070 --> 01:25.010 to 13 at 1:00 a.m. and we can simply do this by passing the timestamp as a string here to the. 01:25.010 --> 01:32.290 Operator And until we can see the temperature information for the specific time stamp here. 01:32.290 --> 01:39.430 So we have here of 1:00 and in L.A. We have ten point seven degrees and in New York minus one point 01:39.430 --> 01:39.880 seven. 01:39.880 --> 01:40.560 So here. 01:41.170 --> 01:47.150 So here we pass an entire timestamp with information on the date and also on the time. 01:47.170 --> 01:48.350 So 1:00. 01:48.460 --> 01:51.840 But we can also pass only to 15. 01:52.000 --> 01:56.660 And by doing so we are selecting all of those are all timestamps in the year to 15. 01:56.660 --> 01:58.450 So let's try this out here. 02:00.700 --> 02:02.430 And here we can see that. 02:02.470 --> 02:10.990 Here we have timestamps to 15 starting from the very first timestamp at the January the 1st at midnight. 02:11.470 --> 02:20.210 And so we have your all time stamps into 15 with hourly increments and uh the very last row we have 02:20.210 --> 02:26.240 here the thirty first of December to 15 at 11:00 p.m.. 02:26.240 --> 02:31.880 So these are all time stamps in the year to 15 and in total we have eight thousand seven hundred and 02:31.880 --> 02:36.730 sixty arose time stamps and we can also slice for particular month. 02:36.770 --> 02:45.190 So for example here for May 215 and here we are getting all the time stamps and the May 215. 02:45.330 --> 02:51.070 Starting from the 1st of May midnight onto the uh thirty first. 02:51.090 --> 02:54.570 Uh 11:00 in the evening. 02:54.570 --> 02:59.040 And in total we have seven hundred forty four rows of timestamps. 02:59.040 --> 03:01.680 And of course we can also select a single day. 03:01.680 --> 03:06.620 So for example here the 20th of uh May 2015 03:10.190 --> 03:13.550 and I think we should get to twenty four rows. 03:13.580 --> 03:15.040 So we have our early data. 03:15.080 --> 03:19.670 And one day so let's check here other save 03:23.670 --> 03:30.350 and we have 24 rows so far each hour on the 20th of May 2015. 03:30.540 --> 03:32.300 And we can also add to your time data. 03:32.310 --> 03:36.210 So if like stop at 10:00 for the 20th of May. 03:38.010 --> 03:39.030 And this also works. 03:39.060 --> 03:41.330 So here this is the one single time stamp. 03:41.850 --> 03:45.750 And in theory we can also add information on the minutes and seconds here. 03:45.750 --> 03:54.020 So for example here we have 10 30 by two we get an error message because we have only hourly data. 03:54.150 --> 03:57.800 So that is no timestamp 10 30. 03:58.050 --> 03:58.870 All right. 03:58.970 --> 04:02.010 And we can also index lost lies for time period. 04:02.030 --> 04:08.550 So for example from January the 1st to 15 until the very last day to 15. 04:08.630 --> 04:11.220 So let's try this out here. 04:11.390 --> 04:15.740 So we are starting here at midnight on January the 1st. 04:15.800 --> 04:22.490 And the last timestamp is at 11:00 at the thirty first. 04:22.490 --> 04:28.970 And actually slicing from the 1st of January and to the thirty first of December it's actually identical 04:28.970 --> 04:31.610 to a slicing only for the year to 15. 04:32.480 --> 04:41.220 So we can check this myth that the equals method and we should get you a true however of course we live 04:41.220 --> 04:42.950 for this life and notation. 04:43.040 --> 04:44.620 We can be more versatile. 04:44.980 --> 04:54.640 So for example we can also slice from the 15th of April to 15 to the twenty third of February to a 16 04:57.030 --> 05:01.530 so obviously in this case we cannot only pass here to 15 after 16 05:04.770 --> 05:13.690 and we can also slice from the 20th of May to 15 and to the very last timestamp so here we are starting 05:13.690 --> 05:18.140 with the 20th of May and the very last timestamp is here. 05:18.430 --> 05:26.110 The very last timestamp in our data frame so the thirty first of December to 60 the 11:00 p.m. 05:29.150 --> 05:32.610 and you can also slice our data frame from the beginning until. 05:32.630 --> 05:36.740 For example um the 20th of May 215. 05:37.010 --> 05:43.400 So here we are starting into 13 and to the 20th of May to 15. 05:43.400 --> 05:48.740 11:00 and same as service for the two daytime method. 05:48.910 --> 05:53.020 Pandas can handle here all kinds of formats. 05:53.050 --> 06:00.640 So for example here we can select for the 20th of February to 15 and we can do this in different formats 06:00.640 --> 06:04.910 here for example in this format and this still works here. 06:06.760 --> 06:16.620 So starting from midnight to 11:00 p.m. However there's one thing that we cannot do. 06:16.740 --> 06:20.430 So let's assume that we want to select several single timestamp. 06:20.430 --> 06:29.960 So for example here the 20th of May 250 and 10:00 and the 20th of May to 50 and 12:00. 06:30.060 --> 06:32.410 And this does not work here. 06:32.550 --> 06:39.210 And in this case first of all we have to transform the two timestamps into a date time index and we 06:39.210 --> 06:43.350 can do this with the method PD got to date time. 06:43.440 --> 06:46.350 So we are passing here a list of the two time stamps. 06:46.390 --> 06:53.860 And let's see what we get here and we are getting it actually the daytime index this hour two time stamps. 06:54.000 --> 06:59.670 And now we can index our data frame that's at the clock operator by passing here the date time index 07:01.380 --> 07:06.160 and by doing so we are getting the two time stamps so 10:00 and twelve o'clock. 07:06.180 --> 07:11.340 And the information for Los Angeles and New York are right. 07:11.340 --> 07:18.250 This is also the video on indexing and slicing time series and with a day time index anti-lock operator. 07:18.350 --> 07:23.510 This is actually quite easy and versatile and with this uh we are finished here and I hope to see you 07:23.510 --> 07:25.430 also in the next video by.