WEBVTT 00:02.630 --> 00:08.450 In this video we will split our data framed by more than one key so in principle we can split our data 00:08.450 --> 00:12.200 friends by one two three or whatever number of keys. 00:12.200 --> 00:14.240 So this is no problem. 00:14.240 --> 00:18.460 And first of all we import pandas as always. 00:18.770 --> 00:25.370 And then now we are working with the Summer Olympics data frame and we will have a look at the first 00:25.370 --> 00:33.620 five rows here and also add some meta information and this time we are working with the whole data frame. 00:34.900 --> 00:37.120 So no toy examples this time. 00:37.120 --> 00:40.900 And let's have a look here at the country column. 00:41.920 --> 00:43.390 So now let's first have a look. 00:43.390 --> 00:48.360 How many different or unique countries we have in our data frame. 00:49.490 --> 00:55.100 So up to now athletes from one hundred forty seven countries have already won medals in some Olympic 00:55.100 --> 01:02.000 Games and in the last video where we have split our Titanic data frame into a major data frame and a 01:02.000 --> 01:08.960 female data frame we have seen that we can also do this as says hello operator and create a group by 01:08.960 --> 01:10.040 group. 01:10.040 --> 01:17.060 So this is quite easy if we only have two groups of males and females but the fear of 147 groups like 01:17.060 --> 01:24.290 in this case then doing this manually with the log operator is getting quite messy and this the group 01:24.290 --> 01:27.100 by a method we can do this actually in one line. 01:27.320 --> 01:28.310 So let's do this. 01:28.310 --> 01:34.520 We are splitting our Summer Olympics data frame by the key country and we're safe. 01:34.520 --> 01:42.650 Our group by object in the Variable split 1 and that's transform our group by up checked into a list 01:43.250 --> 01:45.130 and stored in the variable idea. 01:46.460 --> 01:48.800 So let's have a look here at our list. 01:48.810 --> 01:51.650 Uh first of all let's secure the links. 01:52.070 --> 01:57.680 So we have one of the 47 countries and the length of our list is 147. 01:57.950 --> 02:04.660 And for each country actually we have here in our list a top of the consisting of the country code. 02:04.680 --> 02:10.210 Yes starting with Afghanistan and then the second element in the top. 02:10.220 --> 02:15.950 Actually the data frame with all rows where we have athletes from Afghanistan. 02:15.950 --> 02:20.840 So in total here we have two medals both medals awarded here and tech one though. 02:21.610 --> 02:24.230 And here we have the second element of our list. 02:24.400 --> 02:26.140 So the country H. 02:26.270 --> 02:29.150 Oh so actually I do not know which country it is. 02:29.390 --> 02:32.450 And then the third element is here Algeria. 02:33.540 --> 02:41.040 So we can also randomly select a data frame and we slice for the top of that next position 100 and we 02:41.040 --> 02:44.210 slice for the data frame so let's see what we've got here. 02:47.250 --> 02:55.590 And we get you a data frame with all medals won by athletes from the Philippines and now instead of 02:55.590 --> 03:01.830 splitting our data frame only by the country you want to split our data frame by a country and send 03:01.830 --> 03:10.240 a so give us our group by method we have the parameter by and before we just passed here one column 03:10.240 --> 03:14.980 name but we can pass here a list of multiple column names. 03:14.980 --> 03:22.330 So here we are passing the column name country and gender and by doing so we can split our data frame 03:22.480 --> 03:23.860 by country and gender. 03:24.490 --> 03:29.750 So let's do this and we save our group by object in the variable split 2. 03:30.400 --> 03:37.450 So let's run in the center and then we are transforming our group by object into a list and save the 03:37.450 --> 03:41.230 list and the variable 2 and let's print here L2 03:44.190 --> 03:46.890 and first of all you can see here the first element. 03:47.940 --> 03:53.650 We fear all men from Afghanistan that already won a medal in some Olympic Games. 03:56.220 --> 03:59.320 And we can also check here the length of our list too. 04:00.120 --> 04:08.670 So intuitively we could say OK we have 147 countries and we have male and female saw the number of possible 04:08.670 --> 04:12.480 combinations should be two times 147. 04:12.590 --> 04:19.580 It's 294 So potentially we have two hundred ninety four groups. 04:19.580 --> 04:25.850 But I do think at least for the smaller countries that we have countries where only male athletes are 04:25.850 --> 04:29.330 female at athletes have already won medals. 04:30.020 --> 04:36.240 So I do not expect to have here in our to list a length of 294. 04:36.350 --> 04:37.700 So let's check this out 04:40.560 --> 04:48.220 and we have only two hundred thirty six groups and let's randomly select the group at the next position 04:48.230 --> 04:51.120 one hundred four. 04:51.120 --> 04:57.180 And here we can see that we have the group with female athletes from Ireland. 04:57.270 --> 05:03.690 So actually we have here one two three four five six medals that has been awarded to female athletes 05:03.690 --> 05:05.050 from Ireland. 05:05.220 --> 05:10.840 And apparently here we have a topless so we can also check here for the first element in our topic. 05:12.090 --> 05:15.960 And here it's not a string but also a topic consisting of two strings. 05:15.950 --> 05:23.310 So here are two groups so we have Ireland and women and we can also slice our total for the second element 05:23.310 --> 05:26.310 or the second element here is the data frame. 05:27.780 --> 05:35.790 And here we have the data frame with all rows or medals where we have female athletes from Ireland and 05:35.790 --> 05:37.570 still you on the left side. 05:37.800 --> 05:44.760 We have here the index labels from our original data frame. 05:44.810 --> 05:45.200 All right. 05:45.200 --> 05:46.790 So now on this video we split it. 05:46.790 --> 05:52.120 Our summer Olympics data frame not only by one key but by two keys. 05:52.130 --> 05:55.490 And we could also do the same for our markets. 05:55.490 --> 06:02.000 So for example we could say that we want to split our data frame by Country Athlete and let's say the 06:02.000 --> 06:06.650 year of the addition of the Olympic Games then yeah. 06:06.660 --> 06:09.810 Michelle Marie Smith what actually form one group. 06:09.960 --> 06:17.040 So female athletes from Ireland who won in the addition and in 1960 a medal and a second group would 06:17.040 --> 06:23.280 be a female athletes also from Ireland that won in the year on the addition 2000 a medal. 06:23.550 --> 06:28.850 So you in this data frame we only got have Sonya Sullivan. 06:28.890 --> 06:29.310 All right. 06:29.320 --> 06:33.940 We are finished now with this video and I do think we have now a good impression. 06:33.940 --> 06:40.600 How a group buys splits our original data frame and many data frames on many groups and in the next 06:40.600 --> 06:44.300 video we will learn how split apply combined works. 06:44.320 --> 06:49.300 So this is the real power of group buy and I hope to see on the next video by.