WEBVTT 00:00.930 --> 00:07.050 In this video we are creating several categorical seaborne plots and they all have in common that we 00:07.050 --> 00:09.180 group our data by categories. 00:09.180 --> 00:14.570 So for example in our Titanic data frame we have the categories peak class and sex. 00:14.700 --> 00:20.730 So we split our data into groups and then we compare these groups for example by the H or by the fair 00:21.330 --> 00:24.240 and we can draw conclusions like that. 00:24.240 --> 00:30.630 The first class made a peasant us were older than the asset class female passengers and so on. 00:30.720 --> 00:36.840 So we are still in poverty PANDAS and the seaborne and met plot slip and we still have imparted our 00:36.840 --> 00:44.730 Titanic data frame and for all of our plots we set here the figure size to twelve eight and we increase 00:45.000 --> 00:52.480 the fun scale to a one point five and our very first blood should be a categoric set up a lot of it 00:52.480 --> 00:58.620 is a chipped plot so recall here as an escort to strip plot and that sort of a look here. 00:59.010 --> 01:05.130 So the strip blood actually draws a set up plot where one variable is categorical and also give you 01:05.140 --> 01:11.070 f apparently that data we have we have to pass them our data frame here Titanic and let's assume that 01:11.070 --> 01:12.640 you want to f on the x axis. 01:12.660 --> 01:14.320 Our categorical variables. 01:14.340 --> 01:19.680 So he has sex and on the y axis we want to f the H. 01:20.270 --> 01:22.220 So that's simply around the set here. 01:22.800 --> 01:28.230 So we splitting up our data into male and female A.A for each single male peasant. 01:28.260 --> 01:34.980 We have a dot to the corresponding to the eights and the same we f for females and you can see here 01:35.010 --> 01:37.740 that many points are overlapping. 01:37.740 --> 01:44.220 We feel the pyramid the Gita and by default it's set to false but we can also set this to true 01:47.140 --> 01:52.050 and by doing so we make you the plot of the points that are not overlapping. 01:52.090 --> 02:00.480 So now we can better see that the majority of made a presence us here around the 20 to 35 and we have 02:00.480 --> 02:06.550 some pretty young male peasants us and also some quite old peasant us here and the same we have fear 02:06.550 --> 02:14.980 for the female peasant us and also the strip plot provides for the option to further splitting our two 02:15.310 --> 02:18.070 categories into a more categories. 02:18.070 --> 02:24.370 So instead of having the only male and female we could also have fear may 1st class may 2nd classmates 02:24.380 --> 02:26.970 at class and the same for the females. 02:27.130 --> 02:29.660 And again here we ask for the pyramid a hue. 02:30.100 --> 02:37.580 And by default it said to none party we can pass also a categorical variable or a categorical column. 02:37.840 --> 02:46.700 And for example here we can pass the peak class column so that's running at the center and we can see 02:46.700 --> 02:53.510 here that for each point off each dot we have the additional information about the PE class. 02:53.510 --> 02:55.470 So most of the very old. 02:55.580 --> 02:57.290 Made a peasant Yes I actually blue. 02:57.300 --> 03:04.550 So this is the first class for example but it's actually pretty hard to differentiate between the class 03:04.580 --> 03:10.090 and between the classes and therefore we have the additional parameter dots. 03:10.340 --> 03:13.640 And by default it's to force but we can set it to a true 03:16.680 --> 03:19.030 and by doing so we actually separating here. 03:19.040 --> 03:22.250 The points are the dots for the PE classes. 03:22.350 --> 03:28.710 So now here we have for one line with the first class passenger dots source get us then we have here 03:28.710 --> 03:35.370 one line with the second class mates and so on and actually at a very first glance we can now see here 03:35.490 --> 03:43.260 that the far made us about four females it looks like that first class passengers are older than the 03:43.260 --> 03:47.300 second class passengers and that are older than the third class passengers. 03:47.310 --> 03:55.230 So you can see here that we have here a kind of decreasing trend and we not only can split our data 03:55.260 --> 03:57.930 by the six and the peak class column. 03:58.470 --> 04:02.910 So for example instead of having at the peak class column you can also pass here. 04:02.910 --> 04:07.080 These survived the information to the huge pair in meta 04:10.180 --> 04:18.940 and we have yet in blue all the passengers that did not survive and aren't all passengers that did survive. 04:18.980 --> 04:25.130 All right now let's go on and there is another categorically scatter plot called a swarm plot. 04:25.130 --> 04:26.750 So let's have a look here. 04:26.960 --> 04:32.560 And this one plot actually draws a categorical scatter plot with none overlapping points. 04:33.500 --> 04:40.210 So let's have a look and the swamp blood is pretty much the same as here. 04:40.210 --> 04:43.300 The strip plot with having to eat. 04:43.430 --> 04:44.560 It's true. 04:45.220 --> 04:46.470 So let's have again here. 04:46.480 --> 04:48.280 The PE class to compare it 04:53.060 --> 04:58.290 so these two plots are quite similar but still here with a swamp plot. 04:58.310 --> 05:01.630 We can even better see the distribution here. 05:01.670 --> 05:08.300 So for example here BFA subclass made peasant just that I here all at the same age. 05:08.750 --> 05:15.920 And the swamp plot actually locates all the dots side by side so we can even better see the distribution 05:15.920 --> 05:20.870 or the frequency distribution of the H for the six groups here. 05:21.470 --> 05:26.390 So the most frequent age for asset class made peasant just are here. 05:26.390 --> 05:32.600 And for example for first class my peasant us we can see that there is actually no single concentration 05:32.600 --> 05:33.230 here. 05:33.440 --> 05:39.920 So we have quite a stable distribution across all ages here and that's another plot where we can even 05:39.920 --> 05:43.200 better see the distribution. 05:43.280 --> 05:51.470 And this is the violin plot and again here we f on the x axis and the sex and on the y axis that the 05:51.470 --> 05:59.120 age and additionally we split the male and the female group and two of the three peasant classes. 05:59.720 --> 06:01.070 So let's have a look here. 06:04.010 --> 06:11.060 And instead of having here a setup plot we have here actually a line or a curve where we can see actually 06:11.060 --> 06:13.060 the frequency distribution. 06:13.220 --> 06:15.460 So let's have a look here. 06:16.280 --> 06:21.920 And the line or the curve that you can see here is actually called a kind of density estimate. 06:21.920 --> 06:28.780 So this is kind of a smooth and histogram where we can see actually the frequency distribution slow. 06:28.780 --> 06:29.000 Yeah. 06:29.020 --> 06:35.960 Flux harbor for First Class A made peasant us the highest the frequencies here are around 40 and for 06:35.960 --> 06:43.070 second class peasant just it's around 30 and for a subclass peasant yes that's around 20. 06:43.850 --> 06:47.180 And we can also see here that we have a warning message. 06:47.450 --> 06:49.990 So here a future warning message. 06:50.210 --> 06:56.030 And this has something to do with the interaction between symbol on and the stats library. 06:56.330 --> 07:01.430 So at this point here we can ignore it and actually we can also combine the plot. 07:01.440 --> 07:08.000 So here we have a violent plot and here we have a swarm plot and we can actually combine both plots 07:08.000 --> 07:09.010 into one plot. 07:09.020 --> 07:10.300 So we end Abel here. 07:10.310 --> 07:16.020 These swamp plot and in these swamp plot we have actually hear the same pirate leaders. 07:16.160 --> 07:21.280 So we have on the execs this the sixth column on the right access to the eight. 07:21.740 --> 07:28.180 Then we additionally split up into the peasants classes and in addition we want to have a black collar 07:28.190 --> 07:30.740 for all set us off for dots. 07:30.740 --> 07:31.810 So let's have a look here. 07:34.640 --> 07:40.970 So here we have now a combination of the violent plot and the swarm plot so we can even better see now 07:40.980 --> 07:47.300 that the highest frequency see a flat sample fourth up last May peasant passengers here around the 20th 07:48.200 --> 07:52.680 and for the violin plot there's actually another option here. 07:53.030 --> 07:59.690 So we can see here that on the x axis that we have two categories and for both categories we have sweet 07:59.690 --> 08:03.250 plots but actually a three year swap the lovers here. 08:03.260 --> 08:10.580 So if you FBA on the x axis then the three classes and for each class we have that plot. 08:10.580 --> 08:15.950 So for female and male then we can actually match the two plots. 08:15.980 --> 08:23.840 So yeah again we are creating a violin plot and then we exchange peak class and sex so we pass PE class 08:23.840 --> 08:29.250 to the experimenter and sex to the huge pair. 08:29.420 --> 08:30.850 So let's have a look here. 08:34.140 --> 08:41.470 So here we can see on the x axis the three categories and for each category we have two violin plots. 08:41.490 --> 08:43.750 So for male and female. 08:44.280 --> 08:48.180 And if you have two plots for each category we can also match them. 08:48.180 --> 08:54.450 So we can f you one violin plot and on the left hand side we have actually the distribution father made 08:54.450 --> 08:55.090 a peasant does. 08:55.110 --> 08:58.100 And on the right hand side for the female passengers. 08:58.690 --> 09:03.590 And therefore we have to pass here true to the split pair meet up. 09:04.080 --> 09:07.170 So by default it's set to false but we can set it to true. 09:07.800 --> 09:13.120 And let's see what happens so now here we have only three violins. 09:13.330 --> 09:17.990 So one for the first class one for the second and one for the third class. 09:18.010 --> 09:23.860 So on the left hand side we have the distribution of the HS form male and on the right hand side for 09:23.860 --> 09:26.020 female passengers. 09:26.020 --> 09:30.090 And so we can even better Compare male versus the female. 09:30.140 --> 09:31.330 All right. 09:31.360 --> 09:35.010 And that's in addition to Siebel unplugged called by plot. 09:35.050 --> 09:38.660 And again we pass the Titanic data frame to the perimeter data. 09:38.680 --> 09:45.370 Then on the exact sets we want to have the peak class on the buy access to the age and pass a six to 09:45.370 --> 09:46.840 the Hugh meet us. 09:46.840 --> 09:48.040 So let's have a look here 09:51.330 --> 09:52.950 slow here on the exact source. 09:52.980 --> 10:01.250 We have we peak classless and for each PE class we have female and male or female CRM and male as blue. 10:01.590 --> 10:08.620 And he the buy access that we have for the eights and by default the bomb plot Excelsior calculates. 10:08.670 --> 10:10.680 The mean age for six groups. 10:10.680 --> 10:15.040 So here we are six groups and the height of the Bosnia actually. 10:15.060 --> 10:17.200 The mean age. 10:17.310 --> 10:24.180 So you have a first class made a peasant just we have a mean age of 0 40 then for example for females 10:24.180 --> 10:28.480 here in the first class we have about 34 35. 10:28.620 --> 10:35.190 And these are so-called point estimates because you simply calculated the mean over all the passengers 10:35.220 --> 10:36.920 and the respective group. 10:36.930 --> 10:42.630 And in addition we can see here the black line with skiffs that's kind of a confidence interval. 10:43.050 --> 10:48.990 So instead of having only one point estimate and saying that the average age of first class made a peasant 10:48.990 --> 10:56.970 Justice 42 the confidence interval says that for example in 95 percent of all cases the mean age of 10:57.150 --> 11:02.130 first class passengers should be a B between these two points here. 11:02.140 --> 11:04.380 So this is a more statistical approach. 11:04.660 --> 11:11.740 And with this we can answer the question whether the system historically significant that here the male 11:11.740 --> 11:14.650 passengers are older than the female it hasn't us. 11:14.710 --> 11:21.430 So let's have a look here inside the brackets and we can see here that here we have a confidence interval 11:21.460 --> 11:26.860 of ninety five and we have 1000 simulations. 11:26.920 --> 11:33.760 So seaborne simulates the booking for the Titanic 1000 times and then creates a confidence interval 11:34.100 --> 11:39.760 where we can find for example the mean age of all the first class mail passengers and 95 percent of 11:39.760 --> 11:47.760 all cases so this is see the interpretation and as we can see here that there's hardly any overlapping 11:48.000 --> 11:54.870 so he could conclude that it is statistically significant that the first class made passengers were 11:54.900 --> 11:57.770 older than the first class female passengers. 11:57.870 --> 12:04.440 And for example if we have a look at the second class so here we have an overlapping of the confidence 12:04.440 --> 12:10.240 in the was and yet more hard to say that the difference in the point estimate. 12:10.250 --> 12:10.510 Yes. 12:10.510 --> 12:18.030 So he had the maids a bit older than the females that this is that really is statistically significant. 12:18.030 --> 12:23.970 And instead of having the a bomb plot we can also have the same as a point plot. 12:24.000 --> 12:28.210 So that's the addition of seaborne plot point plot. 12:28.470 --> 12:34.740 And also here we have the same pirate meet us and actually we hear the same plot but instead of having 12:34.740 --> 12:38.350 year pass for each group we have fear for example. 12:38.730 --> 12:44.580 So we ask you one line for the male passengers and we can see here the confidence interval for the mean 12:44.610 --> 12:47.100 age of first class mail passengers. 12:47.100 --> 12:49.550 And this is here the point estimate. 12:49.620 --> 12:54.410 So the mean age of all first class male passengers was this value here. 12:54.510 --> 13:01.620 But if you would repeat the Titanic booking 1000 times that then the mean age is in the 95 percent of 13:01.620 --> 13:02.400 all cases. 13:02.410 --> 13:04.580 And this confidence interval. 13:04.980 --> 13:11.760 And we can see here that there's no overlapping with the female confidence interval and therefore we 13:11.760 --> 13:19.110 could conclude here that the males are older than the female it's of statistical significance. 13:19.260 --> 13:25.770 And the same we could conclude here for the third class but we couldn't conclude this for the second 13:25.770 --> 13:27.590 class. 13:27.600 --> 13:28.070 All right. 13:28.070 --> 13:30.290 This was the the video on categorical plots. 13:30.300 --> 13:32.920 And I hope to see also in the next video by.