WEBVTT 00:01.120 --> 00:06.490 In this video I will provide you with a short summary of this section and also an outlook on the next 00:06.490 --> 00:12.970 section from the last videos we have learned how to import and how to initially inspect a data frame 00:13.270 --> 00:19.840 and also how to slice a select single elements and columns and rows of a data frame and we have also 00:19.840 --> 00:25.210 seen that for most of the tasks and pandas there is more than one way to do the same thing. 00:25.210 --> 00:32.060 However in most cases that's actually a good way to do something and that's a bad way to do something. 00:32.220 --> 00:38.330 And in this video my goal is to highlight and to show you the best practice solutions. 00:38.350 --> 00:44.300 All right so let's start now and whenever we work with pandas we have first of all import pandas as 00:44.310 --> 00:44.890 PD. 00:45.190 --> 00:54.820 So by conventional we import pandas SPDR and we can import data that is stored in a CSP file with the 00:54.880 --> 01:01.960 method PD that read CSA and we have to pass here the file name all right. 01:01.970 --> 01:02.780 So let's do this. 01:02.780 --> 01:06.050 And store here our data frame and the value of the summer 01:08.940 --> 01:15.210 and after having imported the data we need to perform a very first inspection to get a first impression 01:15.220 --> 01:19.360 how the data looks like and therefore we can never look at the very first five rows 01:22.280 --> 01:28.760 and we can also see here the column labels and we can also have a look at the last five rows with the 01:28.760 --> 01:29.540 tail method 01:32.570 --> 01:38.270 and we can call the info method to get to more information more meta information on our data frame 01:40.960 --> 01:47.530 so here we can see that we have over 30000 thousand rows and we have nine columns and we can see here 01:47.620 --> 01:49.840 the data types in each column. 01:49.840 --> 01:50.380 All right. 01:50.410 --> 01:56.260 So now the next step would be to inspect single columns and there are actually two ways to select single 01:56.260 --> 01:56.630 column. 01:56.630 --> 02:00.290 So on the one hand side we have the attribute mutation. 02:00.490 --> 02:09.040 So we can select the athlete column with the top athlete or we can use this map records notation and 02:09.040 --> 02:13.270 pass our column label here athlete within the snap records 02:17.880 --> 02:21.130 so both alternatives gives us actually the same result here. 02:23.270 --> 02:29.450 And the advantage of the attribute notation is actually that you can use the tap completion. 02:29.450 --> 02:37.440 However the attribute notation has some minor disadvantages which we will see later and this cost and 02:37.440 --> 02:45.550 you can also select multiple columns bypassing a list of the column labels within the span brackets 02:45.550 --> 02:45.850 here 02:50.680 --> 02:58.490 so here we have selected the year and the medal column and we can do the same with the log operator 02:58.980 --> 03:01.670 service the key operator we have to pass. 03:01.700 --> 03:07.970 There are labels that you want to select on the left hand side of the comma and the column labels on 03:07.970 --> 03:10.760 the right hand side of the comma here. 03:10.850 --> 03:18.400 And in this case we select all rows and we want to select the year and the middle column and we get 03:18.400 --> 03:19.580 to the same result. 03:19.760 --> 03:26.660 However I always strongly prefer the log notation and the reason for this actually the following. 03:26.660 --> 03:30.170 So if I want additionally select also a row. 03:30.200 --> 03:36.770 So for example let's assume that you want to select the row with the row label one then we can do this 03:36.770 --> 03:39.140 here both in our log operator. 03:39.470 --> 03:41.090 So we can pass here. 03:41.200 --> 03:43.930 They're all labeled one. 03:44.120 --> 03:47.870 And here we get for the medal with the arrow label 1 and the year. 03:47.900 --> 03:54.700 And the medal value so we can actually perform the two indexing operations so we select a row and we 03:54.700 --> 04:02.380 select columns within one indexing operation here at the lock operation and let's go up to our notation 04:02.380 --> 04:02.680 here. 04:02.680 --> 04:08.770 So now if you want to select here additionally the first row then gives you would need to change another 04:08.770 --> 04:11.110 indexing operation to select here. 04:11.140 --> 04:15.140 The row with them the index labeled one So for example we have to change the log. 04:15.150 --> 04:22.770 Operator and select the row with the index label 1 and actually we get to have the same results and 04:22.770 --> 04:23.900 both alternatives. 04:23.970 --> 04:30.490 However applying here the log operator with only 1 indexing operation here is definitely best practice 04:30.510 --> 04:36.570 and more idiomatic and what we did here actually we changed indexing operations so this was called chained 04:36.570 --> 04:41.970 indexing and later and this costs when it comes to manipulating elements in the data frame we will see 04:41.970 --> 04:51.020 that change indexing can lead to a significant confusion and problems so I got always you see the logo. 04:51.020 --> 04:51.480 I. 04:51.500 --> 04:55.970 Operator to select the single elements rows columns and whatever. 04:55.970 --> 04:58.310 So this is very powerful. 04:58.310 --> 05:02.970 All right so let's head on and we can select a rose with the eye log. 05:02.960 --> 05:07.610 Operator And we can perform here positional based indexing. 05:07.940 --> 05:16.860 So let's have a look here so we can select your all rows from an experience from 10 to 20 including 05:18.180 --> 05:21.990 and we can also perform label based indexing. 05:21.990 --> 05:29.570 So first of all let's say we import our summer dataset and we set the athlete column as the index column 05:32.910 --> 05:39.270 and then we can perform here label based indexing with the slope operator and we can pass an indexed 05:39.270 --> 05:42.270 label here within this bad records that we want to have. 05:42.270 --> 05:48.640 So is for example and then we get all instances of convoys. 05:48.660 --> 05:53.790 All right so now let's assume that we want to still all instances of Carl Lewis and we only want to 05:53.790 --> 06:01.630 have the year the event and the medal column and there are many ways to do it. 06:01.720 --> 06:06.280 So peanuts provides us here with full flexibility and many ways to do it. 06:06.280 --> 06:10.010 However as I said before there are good ways and bad ways to do it. 06:10.180 --> 06:11.950 And I show you here two bad ways to do it. 06:11.950 --> 06:14.120 And one good way to do it. 06:14.140 --> 06:19.510 So first of all we could slice our summer data frame by the three columns here. 06:19.510 --> 06:21.650 So let's cross here the rest out. 06:22.540 --> 06:30.250 So first of all we select the the three columns and all rows and then we can chain another index operation. 06:30.250 --> 06:33.100 So here selecting all rows with Carl Lewis 06:37.380 --> 06:42.140 and here we have all rows with conduits and we have the columns year event and medal. 06:42.960 --> 06:47.980 And we can also here rearrange our operations so we take first of all the log. 06:47.970 --> 06:50.850 Operator and select all rows with the conduits 06:55.310 --> 07:02.010 so here we have all rows with Carl Lewis and all columns of our data frame and then be contained and 07:02.010 --> 07:06.510 other indexing operation and can select the year the event and the middle column 07:11.080 --> 07:13.740 so actually here we get the same result as before. 07:13.930 --> 07:19.960 And we actually change to indexing operation so first of all we selected the rows with Carl Lewis and 07:19.960 --> 07:25.320 then we selected the three columns and now the third alternative how to do it. 07:25.330 --> 07:31.210 And the best alternative so I would highly recommend you to do your indexing and slicing operations 07:31.240 --> 07:32.110 in this way. 07:32.440 --> 07:39.370 So we use one indexing operation we use here the local operator and within this one indexing operation 07:39.370 --> 07:45.460 we select all rows with Carl Lewis and we select the columns year event the. 07:45.490 --> 07:51.840 So this is only one indexing operation and we get the same result here. 07:51.840 --> 07:58.260 And even if we do get four all three alternatives the same result the first two alternatives can cause 07:58.260 --> 08:04.740 severe problems later when we try to change and manipulate elements in our data frame because he in 08:04.740 --> 08:08.490 the first two identifies the with the change indexing. 08:08.490 --> 08:11.470 So we chained one index operation after another. 08:11.760 --> 08:15.110 Here we have our first index operation and the second one. 08:15.360 --> 08:23.630 And here we have only one index Operation All right so now let's move on and let's have a look again 08:23.630 --> 08:31.310 at our first five rows and append US library provides us actually with the three different data types 08:31.310 --> 08:37.910 or classes and we already learned that here our summer data frame is a data frame so this is actually 08:37.910 --> 08:42.190 the first the data type and then if we select one column here. 08:42.190 --> 08:48.530 So for example the year column and we can see here that this is a pattern that serious. 08:48.560 --> 08:54.420 So it's a one dimensional labeled array and the data type is here a that serious. 08:55.130 --> 09:00.860 And if you select here the column labels lists the attribute columns. 09:00.860 --> 09:06.980 And if you check your also for the data type then we can see here that this is a panda's index object 09:09.020 --> 09:14.370 and we can also call the ROW Labels with the attribute index. 09:14.570 --> 09:23.670 And this is also an index object before we move on this data frames in the next section we will have 09:23.670 --> 09:27.540 a deeper look into pandas series and pandas index objects. 09:27.570 --> 09:32.800 So these are both very important topics then we later work again with their friends. 09:33.000 --> 09:37.830 So hope you enjoyed this first section and I hope to see you in the next section by.