WEBVTT 00:01.370 --> 00:07.310 In this video we learn how to create matrix plots and in particular heat maps to visualize cross tabular 00:07.310 --> 00:07.820 data. 00:08.150 --> 00:10.870 So we're going to see in a minute what I mean with this. 00:10.880 --> 00:15.550 And first of all we import pandas and seaborne met plot live. 00:16.040 --> 00:22.760 And again we are working with the Titanic data set and we've already learned that we can split our data 00:22.790 --> 00:26.930 into different groups and we can do this by categories. 00:26.930 --> 00:32.570 So for example we have the sixth column and the peak last column and in total we can create six different 00:32.570 --> 00:38.730 groups and then we can determine the size of each group so we can determine how many a first class mail 00:38.750 --> 00:41.930 passengers we had on board of the Titanic. 00:42.110 --> 00:47.570 And we can do this for example with the direct Panda's method the cross tab. 00:47.600 --> 00:51.510 So here we are passing the two categories to sex and peak class. 00:51.530 --> 00:54.880 And by doing so we are getting across tabular data. 00:54.890 --> 00:59.160 So here on the left hand side to the after category is female and male. 00:59.600 --> 01:01.630 And as columns we have two categories. 01:01.690 --> 01:07.220 First second and third class and for each combination of those two categories we FBA. 01:07.250 --> 01:11.630 Actually the amount of passengers that belong to this group. 01:11.900 --> 01:18.020 So for example the largest group by far is in the group made passengers in the third class. 01:18.020 --> 01:24.380 And this is pretty much the section that we need for a matrix plot heat map plot so we can actually 01:24.380 --> 01:31.460 visualize and compare the size of the six groups with a heat map and first flight we determine the figure 01:31.460 --> 01:32.800 size of twelve eight. 01:32.820 --> 01:40.970 And also we want to have a lot of fun scale here and then we are using the as an escort heat map and 01:40.970 --> 01:44.720 to pass our crust tab that we created here. 01:44.750 --> 01:46.820 So let's run here in the cell 01:49.620 --> 01:57.430 so here we f on the x axis the peak class and on the y axis the sex and actually the color indicates 01:57.440 --> 01:59.840 them the size of each group. 01:59.910 --> 02:07.320 So we can see on the right hand side the color bar and dark color means actually around 100 passengers 02:07.430 --> 02:11.860 in this group and the right color means the 300 plus. 02:11.970 --> 02:18.900 And we can also include here the exact size of each group bypassing the true to the m not meter. 02:20.340 --> 02:23.850 So by doing so we're getting the other numbers. 02:23.850 --> 02:27.070 So this will see the largest group Third Class made. 02:27.480 --> 02:31.140 And two we have three hundred forty seven passengers a.. 02:31.200 --> 02:32.640 A bright color. 02:32.640 --> 02:39.950 And that's another pair quite see map and two we can define the color map and actually fly hate map. 02:39.950 --> 02:43.200 The color map red is quite appropriate here. 02:43.560 --> 02:44.510 So let's do this. 02:44.500 --> 02:53.300 What we have here Reds the capital R so here we have red colors and we have dark red for groups that 02:53.310 --> 03:03.230 we have 300 plus passengers and we have bright red for groups that we have around 70 or 100 passengers 03:03.690 --> 03:09.780 and actually we can see here that the scale is not really optimal or perfect yet because here we have 03:09.780 --> 03:11.130 actually one outlier. 03:11.580 --> 03:16.270 So the group a main asset class is by far the largest group. 03:16.470 --> 03:20.670 And the other five groups are pretty much equal here. 03:20.670 --> 03:30.830 So between 76 and 144 and therefore we can cap here our color map for example to 150 03:34.500 --> 03:40.310 so all groups that we have more than one hundred fifty passengers are very dark right here. 03:40.380 --> 03:43.840 And if we can see that we can better differentiate. 03:43.860 --> 03:45.880 The other five groups. 03:45.990 --> 03:50.360 So here we have a very bright color with 76 passengers. 03:50.430 --> 03:54.330 Then we have a darker color with 94 and so on. 03:54.330 --> 03:55.810 So here we have a heat map. 03:55.900 --> 04:01.590 Of the six groups and the color is determined by the size of the groups. 04:01.590 --> 04:06.450 But of course that the color could be also determined by the survival rate. 04:06.450 --> 04:12.600 And again we're using the at the cross tab method and we are splitting our data frame by the sixth column 04:12.600 --> 04:13.850 and the peak last column. 04:14.370 --> 04:20.480 And then we have a look in to the survival column and calculate the mean. 04:20.490 --> 04:24.250 So this is actually the survival rate for all six groups. 04:24.270 --> 04:26.340 So let's calculate this here. 04:26.370 --> 04:33.540 So for example the survival rate for subclass made pairs and justice only 13 percent and the survival 04:33.540 --> 04:38.350 rate for first class female peasant Justice 96 percent. 04:38.370 --> 04:45.390 And again we can pass the cross tabular data to the seaborne heat map and we said to the not pair meter 04:45.390 --> 04:47.900 to true and we want to have four reds. 04:47.910 --> 04:49.080 So let's have a look here. 04:51.830 --> 04:57.670 So here we have our six groups and we can see in dark red color the groups. 04:57.660 --> 05:01.100 That's a high survival rate of around 90 percent. 05:01.100 --> 05:07.250 And and groups with low survival rate are marked with the bright red. 05:07.250 --> 05:11.390 So here we can see the groups made second class and May third class. 05:11.390 --> 05:18.220 Here we have only 40 men 16 percent survival rate and the MFA quite bright color. 05:18.230 --> 05:24.380 All right let's have another example and we can also create a correlation matrix of all numerical features 05:24.380 --> 05:27.280 in our Titanic better frame. 05:27.320 --> 05:32.820 So here we have the correlation matrix and t we have them the animated columns. 05:33.080 --> 05:39.440 So far the correlation to be a failure is between minus 1 and 1 and minus 1 means that perfectly a negative 05:39.440 --> 05:47.110 correlation and plus 1 means perfectly positive correlation and the zero means no correlation. 05:47.270 --> 05:54.110 And we can see that we have a lot of numbers here and it sits at a first glance quite hard to analyze 05:54.110 --> 05:54.350 here. 05:54.350 --> 06:01.490 The correlation matrix and in particular in this case is it makes sense to visualize see the correlation 06:01.490 --> 06:03.780 matrix with a heat map. 06:03.890 --> 06:07.520 So we path a titanic top card to the heat map. 06:07.520 --> 06:12.830 Then again we set an art to true and the color map should be read so again 06:16.090 --> 06:23.500 and here we can see a strong positive correlation in dark red and the negative correlation in bright 06:23.500 --> 06:24.010 color. 06:24.790 --> 06:31.360 So of course in the diagonal we can see perfect positive correlations between the futures itself. 06:31.390 --> 06:33.120 So this is no surprise. 06:33.280 --> 06:36.450 And we can also if I look here at these survived feet. 06:37.120 --> 06:41.220 So we have here a strong negative correlation to the peasant class. 06:41.230 --> 06:46.840 That means first class passengers had a higher chance to survive than a third class passengers. 06:46.850 --> 06:55.660 Then we can see a small negative correlation with the H and the best P features then quite weak positive 06:55.660 --> 07:02.850 correlation with the the pot's feature and a quite significant correlation with the fast feet. 07:03.280 --> 07:08.620 So the higher the ticket price the better the chance to survive the disaster. 07:08.830 --> 07:15.290 And we can see if for example the riders the combination this year between the fair and peak class. 07:15.400 --> 07:22.540 So here we have the most negative correlation and it's no surprise that that between fare and peak class 07:22.620 --> 07:24.850 that's a strong negative correlation. 07:25.180 --> 07:31.720 So passengers in the first class paid more for the tickets than passengers in the third class. 07:31.720 --> 07:38.440 And we can see quite a strong positive correlation between the parts and sip P. 07:38.800 --> 07:46.480 So when passengers traveled with their husband or wife then it was quite likely that also the run also 07:46.480 --> 07:47.460 onboard. 07:47.500 --> 07:53.440 And as a summary we can see here that do we get at our very first glance and intuition on the correlations 07:53.470 --> 07:55.950 without having here a closer look at numbers. 07:55.960 --> 07:58.710 So we can easily see this here graphically. 07:59.260 --> 08:04.900 And this definitely helps if we have fewer even more rows and columns. 08:04.900 --> 08:10.970 So for the human eye some colors are more intuitive than blank numbers here. 08:10.980 --> 08:15.670 All right and with this we are finished here with this video and I hope to see you also in the next 08:15.670 --> 08:16.350 one by.