WEBVTT 00:01.590 --> 00:07.260 Welcome to our first coding video on group by operations in this and the next video. 00:07.260 --> 00:11.640 We want to figure out and really understand what group bite us in the background. 00:11.700 --> 00:15.410 And we also want to understand what a group buy up sectors. 00:15.420 --> 00:20.500 So first of all we will have a deeper look into this split part of split apply and combine. 00:20.730 --> 00:28.620 But first of all let's import pandas and we are working with the Titanic dataset and let's have a look 00:28.620 --> 00:29.820 here at the first five. 00:29.820 --> 00:36.500 Growth in the last five rows and also at some meta information with the Inform method 00:39.810 --> 00:42.950 and for the time being we want to work with a toy example. 00:42.960 --> 00:50.570 So we slice our Titanic data frame only for the first 10 rows and for the column sex and age. 00:50.580 --> 00:54.900 And we are saving here the slice in the variable Titanic slice. 00:54.900 --> 01:00.510 So let's do this and let's have a look here at our slice. 01:00.510 --> 01:03.700 So here we have the first 10 passengers. 01:03.720 --> 01:06.280 And the column sex and age. 01:06.540 --> 01:11.250 And now we want to use the group by method on our slice here. 01:11.250 --> 01:16.110 And what we actually want to do we want to split here our data frame by the sex column. 01:16.120 --> 01:18.450 So the key is the sex column. 01:18.570 --> 01:25.080 And by doing so we expect that we're splitting our data frame and then we get all rows of passengers 01:25.080 --> 01:29.190 who are female and we get all rows and passengers who are male. 01:30.060 --> 01:34.100 So let's also press shift tab within the brackets. 01:34.350 --> 01:40.650 And the most important parameter is here the by parameter where we have to define the keys by which 01:40.650 --> 01:43.050 we want to slice our data frame. 01:43.050 --> 01:46.370 And here we have to pass one or several columns. 01:46.380 --> 01:50.340 And in this case here we are passing the sex column. 01:50.820 --> 01:52.680 So let's do this here. 01:54.210 --> 02:00.380 And here we can see that we get something like a group by object so let's start here. 02:00.390 --> 02:08.840 The group by object in the variable GEO Group by object and let's also check the data type of GPO. 02:09.750 --> 02:18.070 So here it's a data frame grouped by object and same as this all data types are classes in Python. 02:18.110 --> 02:25.450 Also the group by object has specific attributes and methods and one attribute is the group's attributes. 02:25.450 --> 02:26.960 So let's try this out here. 02:28.760 --> 02:32.020 And here we can see that we have two groups and our group by object. 02:32.030 --> 02:36.700 So we have the group female with actually all rows where we have a female. 02:36.710 --> 02:39.990 So we have in the rows 1 2 3 8 9 female. 02:40.010 --> 02:42.760 So let's check this out in our data frame. 02:42.830 --> 02:53.950 1 2 3 8 and 9 so here we have actually the index labels where we have females in our data frame and 02:54.010 --> 02:56.190 the second group is the main a group. 02:56.980 --> 03:04.480 And also here we have the index labels father rows that we have made is that we have here 0 4 5 6 7 03:04.870 --> 03:13.450 0 4 5 6 7 and now we could also transform our group by object into a list by using here the list function 03:13.480 --> 03:22.310 and we save our list and the variable L and let's have a look here at L so it's actually at a first 03:22.310 --> 03:31.320 glance hard to see what we have here but apparently our list has two elements so the first element starts 03:31.320 --> 03:36.560 here with the bracket and ends here and the second element starts and ends here. 03:36.720 --> 03:41.970 So let's check first of all the length of our lists or the amount of elements. 03:41.970 --> 03:47.650 So we have two elements and now let's uh select the first element of our list. 03:49.270 --> 03:54.720 So here we have for the first element and here it looks like a table. 03:54.730 --> 04:00.130 So we have two brackets and apparently our total consists of two elements. 04:00.130 --> 04:01.830 So we have here the string female. 04:02.290 --> 04:05.920 And it seems like that we have here a data frame with all rows. 04:05.950 --> 04:06.970 We have a female. 04:07.900 --> 04:10.150 And we can also check here the data type. 04:10.690 --> 04:17.470 So it should be a total and now we can also select the first element in our total. 04:17.480 --> 04:18.990 So it should be a female. 04:19.550 --> 04:22.920 So let's try this out and it's a female. 04:23.420 --> 04:25.790 And this is apparently a string object. 04:25.940 --> 04:29.540 And let's also select the second element of our total here. 04:29.540 --> 04:31.360 So this looks like a data frame. 04:33.440 --> 04:39.740 And indeed here we have a data frame with all rows where we have females and still here on the left 04:39.740 --> 04:44.270 side the index labels are the same as here and our original data frame. 04:44.870 --> 04:49.900 So here we are for the females the index labels 1 2 3 8 and 9. 04:51.680 --> 04:54.350 And also here we have 1 2 3 8 and 9. 04:55.130 --> 04:56.990 And we can also check out the data types. 04:56.980 --> 05:01.090 So this should be a panda's data frame. 05:01.100 --> 05:05.980 Exactly and for the sake of completeness we can also check in. 05:06.010 --> 05:09.100 The second element and our total here. 05:09.100 --> 05:14.740 So this time we have for a total with the first element the string male and the second element. 05:14.770 --> 05:22.990 Our data frame with rows where we have male passengers. 05:23.050 --> 05:25.640 All right. 05:25.700 --> 05:27.900 So what the group by method actually did. 05:27.920 --> 05:34.550 It's it our original data frame into a two data frames and to the first data frame contains all rows 05:34.580 --> 05:36.030 with female passengers. 05:36.110 --> 05:43.280 And the second data frame contains all rows with the male passengers and alternatively we could also 05:43.280 --> 05:48.960 slice our data frame with the load operator and slice for all rows. 05:49.010 --> 05:50.830 The sex column is female. 05:50.870 --> 05:53.480 So we can also do this on display here. 05:53.810 --> 05:56.930 And here we can see that we get here the same slice. 05:58.040 --> 06:04.760 So here we have all rows with teammates and here we have all rows with females and we can also save 06:04.900 --> 06:08.590 the slice with all females in the variable Titanic slice female. 06:09.490 --> 06:11.920 And we can do the same with male 06:15.670 --> 06:21.910 and then we can check with the equals method whether the data frame was all female passengers which 06:21.910 --> 06:28.780 we created here with the log operator it's actually the same as our slice the group by object here. 06:29.740 --> 06:32.230 So here we get a true. 06:32.740 --> 06:39.610 So this confirms that the group by method splits our data frame by the key is sex and creates two data 06:39.610 --> 06:43.390 frames one female data frame and one male data frame. 06:43.780 --> 06:50.700 And we can also iterate over our group by object with a for loop and for each element in our group by 06:50.700 --> 06:58.050 object we want to print the data frame so we can also do this here and we get here our female data frame 06:58.080 --> 07:01.170 and our male data frame. 07:01.210 --> 07:01.680 All right. 07:01.690 --> 07:09.820 So now let's go also back to our slides here and this is actually exactly what we did right now. 07:10.720 --> 07:15.660 So we created here a group buy up checked and this is actually the split part. 07:15.940 --> 07:21.940 So we had to hear our original data frame or actually our slice of our data frame with nine passengers 07:21.970 --> 07:28.680 and the sex and age column and then we apply to the group by method by the key sex. 07:28.690 --> 07:35.500 And by doing so we split it our data frame into two data frames where in the first data frame we have 07:35.560 --> 07:43.570 all female rows here and in the second data frame we have here all the rows that we have a male passenger. 07:43.990 --> 07:46.060 So this is now our group by object. 07:46.060 --> 07:50.950 So we have two groups two data frames split by the key sex. 07:51.080 --> 07:51.500 All right. 07:51.510 --> 07:57.390 So we are finished now with this video and in the next video we can see that we can also split a data 07:57.390 --> 07:59.330 frame by more than one key. 07:59.340 --> 08:04.260 So by many keys by two or three or four keys so hope to see an error by.