Pass Your Amazon AWS Certified Machine Learning - Specialty Exam Easy!

Amazon AWS Certified Machine Learning - Specialty AWS Certified Machine Learning - Specialty (MLS-C01) exam dumps vce, practice test questions, study guide & video training course to study and pass quickly and easily. Amazon AWS Certified Machine Learning - Specialty AWS Certified Machine Learning - Specialty (MLS-C01) exam dumps & practice test questions and answers. You need avanset vce exam simulator in order to study the Amazon AWS Certified Machine Learning - Specialty certification exam dumps & Amazon AWS Certified Machine Learning - Specialty practice test questions in vce format.

Data Engineering

9. Kinesis Video Streams

Finally, for Kinesis, in the Kinesis family, there is a Kinesis video stream. And so we have the Kinesis video stream right here. And we need producers. So video streaming is just for sending video. And so the producers are what you would expect. It could be a security camera, a body-worn camera, camera deep plans, as Frank will demonstrate, a smartphone camera, audio feeds, images, radar data, or an RTSP camera, which is an image-sending protocol. So this is all the kind of thing I can go into. Can you see our video stream? And then the convention is to have one producer per video stream. So if you have 1,000 cameras, then you will have a thousand video streams. There are video playback capabilities, so you can show the live feed to your applications and your users, and consumers can be a lot of things actually.It could be building your own using MX nets or TensorFlow as machine algorithm frameworks, or it could be using Stage Maker. And Frank will show you this, as well as the Amazon recognition video, which Frank will also show you. You can keep the data in a Kennedy's videostream for up to ten years. So there's lots of data retention that's possible. And this is obviously something you want to do if you have security requirements. For example, if you have a security camera, you definitely want to keep the security camera footage for a long, long time in case something goes wrong. Okay, so more deep-dive architecture about how video streams can be used. And this is from a blog about machine learning, where a Kinesis video stream is being consumed in real time by an application running in a Docker container on target. But that could be easy too as well. It's just an example. That application will be checkpointing the progress of the stream consumption into DynamoDB because it is a consumer. So he wants to make sure that if the Docker container is stopped, then it can retrieve and come back to where it was at the same point of consumption. Then all the frames that have been decodedby the application itself will be sent toSage Maker for doing machine learning based interference. Now, we haven't seen Sage Maker in depth just yet, but Frank will do that for you. But the idea is that it's a machine learning service, and you can get some results out of it. And using these results, we publish all the inference results into a Kinesis data stream that we have created. And that Kinesis data stream can be consumed by a lambda function, for example, to get notification in real time. So, with this architecture, we can apply machine learning algorithms directly in realtime to a video stream and send and convert this into tangible actionable data, into a cancer data stream, so that our other applications can consume that stream and perform whatever needed, such as notifications, which could be very useful. For example, if you want to detect a burglar in your house or someone that's not used to being in your house, this is the kind of architecture you may want to use in AWS to achieve that in AWS.Okay, so that's it for Kennedy's video stream. I think it's very simple to go to the exam. Just think of video streams, really. And I will see you at the next lecture.

10. Kinesis ML Summary

So this was a long section on Kinesis, but here's a quick summary. You need to use Kinesis data streams when you want to create real-time machine learning applications. Okay? For example, if you want to evaluate a stream of data against a Sage Maker service, or if you want to use KinesisData firehose, this will be used when you want toingest a massive amount of data in near real time, put it into S3, and then possibly later perform some training on an algorithm to train your machine learning models. Then can you use data analytics is if youwant to do realtime ETL extract, transform, or load,or you want to do real time ML algorithm. We've seen the random cuts and the hotspots on streams of data. So, as you can see, data analytics is when we don't want to code a lot, we want to use SQL, and we want to apply some quick and easy algorithms to our streams to do some analysis. Finally, we have the Kinesis video stream, which is a real-time video stream against which we would apply machine learning algorithms such as detecting plate numbers, detecting faces, and so on to create machine learning applications in real time. Okay, so hopefully that makes a lot of sense now going tothe exam and I will see you in the next lecture.

11. Glue Data Catalog & Crawlers

So now let's talk about the GlueData catalogue and the Glue crawlers. Glue Data Catalog is a metadata repository for all your tables within your account. So what it will have is that it will have automated schema interference for all these tables, and all the schemas will be versioned. So the idea is that you'll have your "Glue Data Catalog" and will index, for example, all your data sets within Amazon S 3. The really cool thing is that this metadata repository now has schema and data discovery integration with Athena, or the richest spectrum. So the idea is that you can use the schemas from the Glue Data Catalog and import them into your favourite data warehouse or your favourite serverless sequel query tools. And the glue crawlers, as we'll see in this lecture, can help you build the glue data catalog. And the crawlers will just go around all your datatables and your databases and your S3 buckets and figure out what data is there for you. So let's talk about these crawlers in details. So they will go, for example, through S 3, and they will infer all the schemas and the partitions. And it works for Jason's Park CSV relational data store. There are a lot of different data types, and it works for different data sources. So you can have a crawler on Amazon S3, Amazon Redshift, Amazon RDS, and so on. The crawlers can be run on theschedule so regularly or on demand. And you need an impersonation role to give to the crawler to access the data store or credentials, for example, if it wants to access Redshift or RDS. So Glue also has the concept of partitions. So Glue Crawlers will extract the partitions based on how your data in Essex is organized. And so you need to think up front about how you will be organising data because based on that, the partition will be defined and the queries will be optimized. So, for example, you send data from the device—sensor data—and it sends it every hour. Do you query primarily by date ranges? So you first want to know the date that happened in the past hour for all my devices. If so, you may want to organise your buckets so that you have the year, the month, and the day before the device ID. But if you query primarily by device, so you want to give the users the ability to look up their device and then look up in time, then you may want to organise your S revocates to have the device first and then finally the year, month, and date. So this gives you two different kinds of partitioning schemes in S Three, and that is very important when you go ahead and run your queries on it afterwards. Okay, so let's go. In the next lecture, I will be creating our first Glue Data Catalog and our first Glue Crawler to look at the data sets. We look forward to seeing you at the next lecture.

12. Lab 1.3 - Glue Data Catalog

Okay, so now we are going to go to the glue service. And with the glue service, we are going to create our first crawler and our first data store. So I'm going to go to the crawlers on the left-hand side and I'm going to add a crawler. And I'll call this one Democrat Crawler No. 3. I'll click on Next, and it will connect to a data store. And click on "next." And the data store is going to be S Three. Now I need to specify the path of the data store. And it's going to be the S-three bucket. So, S Three, and then the name of the S Three bucket is right here. So I'll just copy this entire bucketname and paste it in here. Okay. And then I need to add a trailing slash to make sure that everything is included. So I'll click on Next, and do I want to add another data store? No, this is fine. Then do I want to create an IM role for this? So yes, my crawler will need to have the capability to look up what's in my Amazon S-3 buckets. So I'll make an im role that is a glue service role demo. Okay, click on Next, and sometimes there is a bug. So it says that the role already exists. So it was being created, but somehow the UI didn't go through. So I select an ImRole and then refresh this page. and click on Demo. And here we go. Click on "next." The frequency is run on demand, but we could setup a schedule for hourly, daily, Tuesdays, weekly, and so on. And then finally, the database. We're going to add this to our machine learning database. So I'm going to create a new database called Machine Learning. And then the prefix added to the tables is optional. We could set a prefix. For example, I could say "S Three," and all the tables will be prefixed by "S Three." This is just a way of doing it, but for now, I won't need it. I'll just click on Next and be done. Finally, I clicked on Finish. And here we go. So the demo, the crawler, has been created now. Do we want to run it now? Yes, run it now. And now the crawler is going to run. and it's going to go over all my files in my directories in this bucket to figure out what's in it. So let's wait for the crawler to be done, and I'll get back to you. So my crawler has now run, and three tables were added. So if I go to my databases and click on "Machine Learning," I can see and click on tables in my database. And here are the three tables that were found by my crawler. As we can see, they correspond tothe three directories we had in here. So we have the instructor table, and it is a CSV data set. Then we have the ticker analytics, which is Jason, and then we have the ticker demo, which is Jason as well. So if I click on "ticker demo," as we can see, we have the location, and in the bottom, it has figured out the schema for me. So the real cool thing is that it found thatticker symbol sector change and price were columns, but alsoit has figured out that I have partitions, partition zero,partition one, partition two and partition three and these correspondto the directories are here, partition zero, one, two andthree and then the data is there. So the really cool thing is that the groupcrawler was able to detect all the partitioning schemes as well, and I can view the partitions, and it shows me exactly the data in the partitions. So I can see 2019-10-23, 10. This is excellent, and I can close the partitions, or I could go ahead and edit the schema if I wanted to, for example by editing the name. So this is year, this is month,this is day and this is our. So now I've named my partitions, and this is really cool because now I can do better queries against my data, so I can do the exact same thing on my ticker analytics, for example. And we have again four partitions, so I can click on "edit schema," and again I could say "year" and then "month." day and hour, and this will be extremely helpful when we start running queries. So again, I save, and then I can view the partitions, and yes, everything looks good. I can close the partitions and be done. So this is a quick and easy way. And then finally for instructors, as we can see, we have three partitions, and it has even figured out the column names directly because I had a header in there. So really really cool. Okay, so that's it. We have a crawler, and it was crawling three databases, and we have tables in our database. This will be extremely helpful when we get to querying our data. But that's it for this lecture, congratulations andI will see you in the next lecture.

13. Glue ETL

So, finally, there is glue ETL. And Glue ETL stands for Extract, Transform, and Load, which definitely allows you to transform data into clean data and rich data before you do an analysis or train a machine learning model on it. So the idea is that with Blue ETL, you can generate ETL code either in Python or in Scala, and you can modify the code directly, or you can provide your own Spark or Spice Park script directly. And the destinations of your Glue ETL can be S3, JDBC for RDS and Redshift, or the glue data catalog, for example. So the really cool thing about Glue ETL is that it is fully managed, it is cost-effective, and you only pay for the resources consumed. So if you have some Spark jobs, you can have all your jobs run on a serverless Spark platform. And you don't need to worry about provisioning a Spark cluster; you don't need to worry about how to run your jobs. ETL does that for you. So I think in that way, it is quite a cool framework and service. Now, you can use the glue scheduler to schedule jobs, and you can also have glue triggers to automate job runs based on events that happen. So, before you take the exam, keep in mind that glue ETL provides some transformation. Transformations can be bundled; for example, drop fields or drop null fields. And I think the name is pretty obvious: it removes fields (or it will remove null fields) and then filters. For example, if you want to specify a function to filter records, join to enrich the data map to add fields, delete fields, or perform action lookups, And then there's one thing that you remember, because this is a machine learning exam we're talking about. There's something called machine learning transformations, and there's only one right now, and it's called fan match machine learning. And this one can be used to identify duplicates or matching records in your data set even when the records don't match exactly. So this is a really cool thing because with this finding matches and learning transform, you can do some deduplication in your Glue ETL, and this is something that Glue ETL offers as part of that; you can finally do some format conversions. So you can convert between CSV, JSON, avoid parking A, cXML, and so on. And you can use any Apache Spark transformation, for example, the Kmeans algorithm. So let's go into the Glue ETLUI to see how things work.

Go to testing centre with ease on our mind when you use Amazon AWS Certified Machine Learning - Specialty vce exam dumps, practice test questions and answers. Amazon AWS Certified Machine Learning - Specialty AWS Certified Machine Learning - Specialty (MLS-C01) certification practice test questions and answers, study guide, exam dumps and video training course in vce format to help you study with ease. Prepare with confidence and study using Amazon AWS Certified Machine Learning - Specialty exam dumps & practice test questions and answers vce from ExamCollection.