WEBVTT 00:01.240 --> 00:01.840 Hi everyone. 00:01.840 --> 00:02.560 Welcome back. 00:02.560 --> 00:03.320 In this lecture. 00:03.320 --> 00:07.000 Let's understand the windows sliding gatherer in detail. 00:07.240 --> 00:13.920 This gatherer is quite interesting because it helps us create overlapping groups or sliding windows 00:13.920 --> 00:15.960 over the elements of the stream. 00:15.960 --> 00:23.240 When we say overlapping windows, it means that each new window shares one or more elements with a previous 00:23.240 --> 00:23.600 one. 00:24.080 --> 00:30.480 So unlike windows fixed, which forms independent, non-overlapping batches, windows slide move the 00:30.480 --> 00:36.360 window forward by one element at a time, keeping part of the previous window. 00:36.560 --> 00:42.040 This allows us to continuously observe how the data evolves element by element. 00:42.040 --> 00:47.600 For example, in this case we have a stream of four elements A, B, C, and d. 00:47.800 --> 00:54.800 If we apply windows sliding to, this will produce the following windows AB, BC and CD. 00:55.040 --> 01:02.080 You can see that from the first and second window, you have a B element that is being shared in both 01:02.080 --> 01:02.760 the windows. 01:02.800 --> 01:04.720 This is called overlapping windows. 01:04.960 --> 01:13.200 This concept is achieved using the windows sliding gatherer, so this behavior is extremely useful when 01:13.200 --> 01:20.160 you want to analyze sequences or patterns where context matters, not just individual elements. 01:20.160 --> 01:26.640 For instance, if you're analyzing movie ratings over time, you can take every 2 or 3 consecutive movies 01:26.640 --> 01:29.280 and calculate their average rating. 01:29.280 --> 01:33.120 This gives you a sense of how ratings are changed gradually. 01:33.240 --> 01:37.680 Instead of just looking at the movie, which is one movie at a time. 01:38.120 --> 01:43.480 In summary, Windows Sliding helps us process data continuously with overlapping groups. 01:43.480 --> 01:50.240 This is perfect for real time analytics, running averages or sequential comparisons in streams. 01:50.240 --> 01:53.560 Let's go back to the code and explore this particular functionality. 01:55.520 --> 01:58.200 So I'm back in the Stream Gatherer demo class. 01:58.200 --> 02:03.360 So in this case I'm going to comment out the windows sliding function call so that we can actually look 02:03.400 --> 02:06.280 at the data of windows sliding in detail. 02:06.320 --> 02:07.640 So let's go back to this one. 02:07.680 --> 02:09.280 It's going to be the same concept. 02:09.360 --> 02:13.800 So in this case we are going to be doing movie stream dot stream. 02:14.000 --> 02:19.680 And then we are going to be limiting to only five movies because we have 12 movies in the collection. 02:19.840 --> 02:24.400 So using five will help you look at the data in a much clearer way. 02:24.440 --> 02:27.560 So in this case I'm going to be doing the dot gather. 02:27.560 --> 02:33.080 So I'm going to be calling the gatherers dot windows sliding okay. 02:33.120 --> 02:38.880 And then I'm going to be giving the sliding window as two value so that you know what movies are going 02:38.880 --> 02:40.560 to be part of each and every window. 02:40.640 --> 02:42.720 So and then I'm going to be printing this value. 02:42.720 --> 02:46.600 So I'm going to change this one to be a sliding window. 02:46.720 --> 02:48.560 So in this case we have a stream of movie. 02:48.560 --> 02:53.960 As you can see when it when this one happens you have a stream list of movies of size two. 02:54.320 --> 02:56.200 Let me rename this one to be Sliding Window. 02:56.240 --> 02:56.880 There you go. 02:57.000 --> 03:02.280 And then we are going to be accessing those two movies and then printing them over here. 03:02.520 --> 03:02.920 Okay. 03:03.080 --> 03:08.290 Now let's execute this function and then look at the result to get a clearer picture of what this sliding 03:08.290 --> 03:09.450 window creates. 03:09.610 --> 03:12.210 Go to the top and then execute this function. 03:12.250 --> 03:12.570 There you go. 03:12.610 --> 03:15.490 So now you can see that the sliding window is created. 03:15.530 --> 03:19.010 As you can see, we have limited the movie to five right. 03:19.050 --> 03:20.210 Let's go back to the code. 03:20.250 --> 03:22.170 So in here we are limited that to five. 03:22.410 --> 03:24.970 So basically how many windows it created. 03:24.970 --> 03:27.450 1234. 03:27.650 --> 03:28.170 Right. 03:28.210 --> 03:31.330 And if you take a look at it this is the first window which is The Godfather. 03:31.530 --> 03:33.610 And then Shawshank Redemption. 03:33.610 --> 03:37.410 In the second window you see that same movie that is being present. 03:37.570 --> 03:41.090 The last movie in the previous window is a first element in the next window. 03:41.250 --> 03:44.690 And then we have the pulp function, and the same is being repeated. 03:44.690 --> 03:47.970 The last movie in the previous window is the first element in the next window. 03:48.090 --> 03:51.290 So in this case let's go ahead and remove this limit. 03:51.410 --> 03:51.730 Okay. 03:51.770 --> 03:53.850 And then I'm going to change this to four. 03:54.170 --> 03:55.330 Let's see what happens. 03:55.530 --> 03:57.050 Let's rerun the program again. 03:57.210 --> 03:57.410 Okay. 03:57.450 --> 03:58.050 So there you go. 03:58.090 --> 04:00.890 We have this whole sliding window created. 04:01.010 --> 04:04.890 So in this case so you have the first window with four elements. 04:04.890 --> 04:07.970 In the next window you can see that the top. 04:08.250 --> 04:14.050 The last three elements are the first three elements in the next window, so it's always the first element 04:14.050 --> 04:15.890 in the first window is not included. 04:16.050 --> 04:18.090 The rest of the elements are going to be part of this one. 04:18.090 --> 04:19.770 The same concept is repeated. 04:19.810 --> 04:23.850 The last three in the top window is the first three in the next window. 04:23.850 --> 04:30.450 So this concept is called window sliding, where we create overlapping windows of specific size. 04:30.690 --> 04:34.810 And each window shares n elements with a window. 04:34.810 --> 04:42.610 So that means it's going to be sharing the first elements or first windows N elements to the next window. 04:42.610 --> 04:44.650 So in this case if you have to think about it. 04:44.810 --> 04:46.810 So here we have four elements right. 04:46.810 --> 04:51.250 And the next window is going to be holding on n minus one which is four minus one. 04:51.250 --> 04:52.690 In this case the three elements. 04:52.690 --> 04:54.370 That's what we see over here. 04:54.370 --> 04:57.690 So this is all about the window sliding gatherer. 04:57.810 --> 05:04.530 If you have a use case where you have to do a trend analysis, window sliding is a very handy operator 05:04.530 --> 05:06.170 to use for that use case. 05:06.170 --> 05:07.610 This marks the end of this lecture. 05:07.610 --> 05:08.610 Thank you for watching.