1648693800

Compare expected probabilities to what really happens when we run experiments.

- Let's say that you've got a bag, and in that bag you put a bunch of marbles. So, let's say you put 50 of these magenta marbles. So one, two, three, four, five, six, seven, I'm not gonna draw all of them but you get the general idea. There are going to be 50 magenta marbles, and there's also going to be 50 blue marbles. And what you do is, you have these 100 marbles in there, half of them magenta, half of them blue. And before picking a marble out, and you're gonna be blindfolded when you pick a marble out, you shake the bag really good to, so you think, mix them up a little bit. And so, if you were to say theoretically, what is the probability, if you stuck your hand in and you're not looking, what is the probability of picking a magenta? I feel the need to write magenta in magenta. What is the probability of picking a magenta marble? Well, theoretically there's 100 equally likely possibilities, there's 100 marbles in the bag. And 50 of them involve picking a magenta. So, 50 out of 100, when this is the same thing as a 1/2 probability. So you could say, well, "Theoretically, "there is a 1/2 probability, I just did the math." If you say these are 100 equally likely possibilities, 50 of them are picking magenta. Now let's say you actually start doing the experiment. So you literally take a bag with 50 magenta marbles, 50 blue marbles, and then you start picking the marbles, and then you see what marble color you picked, and you put it back in, and then you do it again. And so, let's say that after every time you put your hand in the bag and you take something out of the bag, and you observe what it is, we're gonna call that an experiment. So, after 10 experiments, let's say that you you have picked out seven magenta and three blue. So, is this strange that out of the first 10 experiments, you haven't picked out exactly half of them being magenta, you've picked out seven magenta, and then the other three were blue. Well no, this is definitely a reasonable thing. If the true probability of picking out a magenta is 1/2, it's definitely possible that you could still pick out seven magenta, that just happened to be what your fingers touched. And this isn't a lot of experiments, it's completely reasonable that out of 10, yeah, you could have, later on in statistics we'll define these things in more detail, but there's enough variation in where you might pick that you're not going to always get, especially with only 10 experiments, you're not definitely going to get exactly 1/2. Instead of having five magenta, it's completely reasonable to have seven magenta. So, this really wouldn't cause me a lot of pause. I still wouldn't question what I did here when I calculated this theoretical probability. But let's say you have a lot of time on your hands. And let's say after 10,000 trials here, after 10,000 experiments, and remember the experiment; you're sticking your hand in the bag without looking, your fingers kind of feeling around, picks out a marble, and you observe the marble and you record what you found. And so, let's say after 10,000 experiments, you get 7,000 magenta. Actually I'm gonna do slightly different numbers, so let me make it even more extreme. Let's say you get 8,000 magenta and you have 2,000 blue. Now this is interesting, because here what you're seeing experimentally seems to be very different. And now you have a large number of trials right over here, not just 10. 10 is completely reasonable that, hey you know, I got seven magenta and three blue instead of five and five, but now you've done 10,000. You would've expected if this was the true probability, you would've expected that half of these would've been magenta, only 5,000 magenta and 5,000 blue, but you got 8,000 magenta. Now, this is within the realm of possibility if the true probability of picking a magenta is 1/2, but it's very unlikely that you would've gotten this result with this many experiments, this many trials if the true probability was 1/2. Here your experimental probability is showing, look, out of 10,000 trials, experimental probability here is you had 10,000 trials, or 10,000 experiments I guess you could say. And and in 8,000 of them, you got a magenta marble. And so, this is going to be 80%, or 8/10. So, there seems to be a difference here. The reason why I would take this more seriously is that you had a lot of trials here, you did this 10,000 times. If the true probability was one half, it's very low likelihood that you would've gotten this many magenta. So, when you think about it you're like, "What's going on here, what are "possible explanations for this?" This, I wouldn't have fretted about, after 10 experiments, not a big deal. But after 10,000, this would have caused me pause. Well, why would this happen, I mixed up the bag every time. And there're some different possibilities; maybe the blue marbles are slightly heavier, and so when you shake the bag up enough, the blue marbles settled to the bottom, and you're more likely to pick a magenta marble. Maybe the blue marbles have a slightly different texture to them, in which case, maybe they slip out of your hands, or they're less likely to be gripped on, and so you're more likely to pick a magenta. So, I don't know the explanation, I don't know what's going on in that bag, but if I thought theoretically that the probability should be 1/2, because half of the marbles are magenta, but I'm seeing through my experiments that 80% of what I'm picking out, especially if I did 10,000 of them, if I did this 10,000 times, well, this is going to cause me some pause. I would really start to think about whether it's truly equally likely for me to pick out a red, a magenta versus a blue. Something else must be going on.

1648693800

Compare expected probabilities to what really happens when we run experiments.

- Let's say that you've got a bag, and in that bag you put a bunch of marbles. So, let's say you put 50 of these magenta marbles. So one, two, three, four, five, six, seven, I'm not gonna draw all of them but you get the general idea. There are going to be 50 magenta marbles, and there's also going to be 50 blue marbles. And what you do is, you have these 100 marbles in there, half of them magenta, half of them blue. And before picking a marble out, and you're gonna be blindfolded when you pick a marble out, you shake the bag really good to, so you think, mix them up a little bit. And so, if you were to say theoretically, what is the probability, if you stuck your hand in and you're not looking, what is the probability of picking a magenta? I feel the need to write magenta in magenta. What is the probability of picking a magenta marble? Well, theoretically there's 100 equally likely possibilities, there's 100 marbles in the bag. And 50 of them involve picking a magenta. So, 50 out of 100, when this is the same thing as a 1/2 probability. So you could say, well, "Theoretically, "there is a 1/2 probability, I just did the math." If you say these are 100 equally likely possibilities, 50 of them are picking magenta. Now let's say you actually start doing the experiment. So you literally take a bag with 50 magenta marbles, 50 blue marbles, and then you start picking the marbles, and then you see what marble color you picked, and you put it back in, and then you do it again. And so, let's say that after every time you put your hand in the bag and you take something out of the bag, and you observe what it is, we're gonna call that an experiment. So, after 10 experiments, let's say that you you have picked out seven magenta and three blue. So, is this strange that out of the first 10 experiments, you haven't picked out exactly half of them being magenta, you've picked out seven magenta, and then the other three were blue. Well no, this is definitely a reasonable thing. If the true probability of picking out a magenta is 1/2, it's definitely possible that you could still pick out seven magenta, that just happened to be what your fingers touched. And this isn't a lot of experiments, it's completely reasonable that out of 10, yeah, you could have, later on in statistics we'll define these things in more detail, but there's enough variation in where you might pick that you're not going to always get, especially with only 10 experiments, you're not definitely going to get exactly 1/2. Instead of having five magenta, it's completely reasonable to have seven magenta. So, this really wouldn't cause me a lot of pause. I still wouldn't question what I did here when I calculated this theoretical probability. But let's say you have a lot of time on your hands. And let's say after 10,000 trials here, after 10,000 experiments, and remember the experiment; you're sticking your hand in the bag without looking, your fingers kind of feeling around, picks out a marble, and you observe the marble and you record what you found. And so, let's say after 10,000 experiments, you get 7,000 magenta. Actually I'm gonna do slightly different numbers, so let me make it even more extreme. Let's say you get 8,000 magenta and you have 2,000 blue. Now this is interesting, because here what you're seeing experimentally seems to be very different. And now you have a large number of trials right over here, not just 10. 10 is completely reasonable that, hey you know, I got seven magenta and three blue instead of five and five, but now you've done 10,000. You would've expected if this was the true probability, you would've expected that half of these would've been magenta, only 5,000 magenta and 5,000 blue, but you got 8,000 magenta. Now, this is within the realm of possibility if the true probability of picking a magenta is 1/2, but it's very unlikely that you would've gotten this result with this many experiments, this many trials if the true probability was 1/2. Here your experimental probability is showing, look, out of 10,000 trials, experimental probability here is you had 10,000 trials, or 10,000 experiments I guess you could say. And and in 8,000 of them, you got a magenta marble. And so, this is going to be 80%, or 8/10. So, there seems to be a difference here. The reason why I would take this more seriously is that you had a lot of trials here, you did this 10,000 times. If the true probability was one half, it's very low likelihood that you would've gotten this many magenta. So, when you think about it you're like, "What's going on here, what are "possible explanations for this?" This, I wouldn't have fretted about, after 10 experiments, not a big deal. But after 10,000, this would have caused me pause. Well, why would this happen, I mixed up the bag every time. And there're some different possibilities; maybe the blue marbles are slightly heavier, and so when you shake the bag up enough, the blue marbles settled to the bottom, and you're more likely to pick a magenta marble. Maybe the blue marbles have a slightly different texture to them, in which case, maybe they slip out of your hands, or they're less likely to be gripped on, and so you're more likely to pick a magenta. So, I don't know the explanation, I don't know what's going on in that bag, but if I thought theoretically that the probability should be 1/2, because half of the marbles are magenta, but I'm seeing through my experiments that 80% of what I'm picking out, especially if I did 10,000 of them, if I did this 10,000 times, well, this is going to cause me some pause. I would really start to think about whether it's truly equally likely for me to pick out a red, a magenta versus a blue. Something else must be going on.

1593242571

Techtutorials tell you the best online IT courses/training, tutorials, certification courses, and syllabus from beginners to advanced level on the latest technologies recommended by Programming Community through video-based, book, free, paid, Real-time Experience, etc.

#techtutorials #online it courses #mobile app development courses #web development courses #online courses for beginners #advanced online courses

1672736040

Our MLOps Zoomcamp course

Teach practical aspects of productionizing ML services — from collecting requirements to model deployment and monitoring.

Data scientists and ML engineers. Also software engineers and data engineers interested in learning about putting ML in production.

- Python
- Docker
- Being comfortable with command line
- Prior exposure to machine learning (at work or from other courses, e.g. from ML Zoomcamp)
- Prior programming experience (at least 1+ year)

Course start: 16 of May

The best way to get support is to use DataTalks.Club's Slack. Join the `#course-mlops-zoomcamp`

channel.

To make discussions in Slack more organized:

- Follow these recommendations when asking for help
- Read the DataTalks.Club community guidelines

- What is MLOps
- MLOps maturity model
- Running example: NY Taxi trips dataset
- Why do we need MLOps
- Course overview
- Environment preparation
- Homework

- Experiment tracking intro
- Getting started with MLflow
- Experiment tracking with MLflow
- Saving and loading models with MLflow
- Model registry
- MLflow in practice
- Homework

- Workflow orchestration
- Prefect 2.0
- Turning a notebook into a pipeline
- Deployment of Prefect flow
- Homework

- Three ways of model deployment: Online (web and streaming) and offline (batch)
- Web service: model deployment with Flask
- Streaming: consuming events with AWS Kinesis and Lambda
- Batch: scoring data offline
- Homework

- Monitoring ML-based services
- Monitoring web services with Prometheus, Evidently, and Grafana
- Monitoring batch jobs with Prefect, MongoDB, and Evidently

- Testing: unit, integration
- Python: linting and formatting
- Pre-commit hooks and makefiles
- CI/CD (Github Actions)
- Infrastructure as code (Terraform)
- Homework

- End-to-end project with all the things above

- CRISP-DM, CRISP-ML
- ML Canvas
- Data Landscape canvas
- MLOps Stack Canvas
- Documentation practices in ML projects (Model Cards Toolkit)

(In October)

- Larysa Visengeriyeva
- Cristian Martinez
- Kevin Kho
- Theofilos Papapanagiotou
- Alexey Grigorev
- Emeli Dral
- Sejal Vaidya

- Machine Learning Zoomcamp - free 4-month course about ML Engineering
- Data Engineering Zoomcamp - free 9-week course about Data Engineering

**I want to start preparing for the course. What can I do?**

If you haven't used Flask or Docker

- Check Module 5 from ML Zoomcamp
- The section about Docker from Data Engineering Zoomcamp could also be useful

If you have no previous experience with ML

- Check Module 1 from ML Zoomcamp for an overview
- Module 3 will also be helpful if you want to learn Scikit-Learn (we'll use it in this course)
- We'll also use XGBoost. You don't have to know it well, but if you want to learn more about it, refer to module 6 of ML Zoomcamp

**I registered but haven't received an invite link. Is it normal?**

Yes, we haven't automated it. You'll get a mail from us eventually, don't worry.

If you want to make sure you don't miss anything:

- Register in our Slack and join the
`#course-mlops-zoomcamp`

channel - Subscribe to our YouTube channel

**Is it going to be live?**

No and yes. There will be two parts:

- Lectures: Pre-recorded, you can watch them when it's convenient for you.
- Office hours: Live on Mondays (17:00 CET), but recorded, so you can watch later.

**I just joined. Can I still get a certificate?**

- To get a certificate, you need to complete a project
- There will be two attempts to do a project
- First: in July, second: in August
- If you manage to finish all the materials till August, and successfully finish the project, you'll get the certificate

- Sign up here: https://airtable.com/shrCb8y6eTbPKwSTL (it's not automated, you will not receive an email immediately after filling in the form)
- Register in DataTalks.Club's Slack
- Join the
`#course-mlops-zoomcamp`

channel - Tweet about the course!
- Subscribe to the public Google calendar (subscription works from desktop only)
- Start watching course videos! Course playlist
- Technical FAQ
- For announcements, join our Telegram channel
- Leaderboard

Author: DataTalksClub

Source Code: https://github.com/DataTalksClub/mlops-zoomcamp

1648726500

In this Precalculus Course's video, we will learn the Experimental versus theoretical probability simulation.

What we're going to do in this video is explore how experimental probability should get closer and closer to theoretical probability as we conduct more and more experiments or as we conduct more and more trials. This is often referred to as The Law of Large Numbers. If we only have a few experiments, it's very possible that our experimental probability could be different than our theoretical probability or even very different. But as we have many many more experiments, thousands, millions, billions of experiments, the probability that the experimental and the theoretical probabilities are very different, goes down dramatically. But let's get an intuitive sense for it. This right over here is a simulation created by Macmillan USA. I'll provide the link as an annotation. And what it does is it allows us to simulate many coin flips and figure out the proportion that are heads. So right over here, we can decide if we want our coin to be fair or not. Right now, it says that we have a 50% probability of getting heads. We can make it unfair by changing this but I'll stick with the 50% probability. If we wanna show that on this graph here, we can plot it. And what this says is at a time, how many tosses do we wanna take. So let's say, let's just start with 10 tosses. So what this is going to do is take 10 simulated flips of coins with each one having a 50% chance of being heads. And then as we flip, we're gonna see our total proportion that are heads. So let's just talk through this together. So starting to toss. And so what's going on here after 10 flips? So as you see, the first flip actually came out heads and if you wanted to, say what's your experimental probability after that one flip, you'd say well, with only one experiment, I got one heads so it looks like 100% were heads. But in the second flip, it looks like it was a tails. Because now the proportion that was heads after two flips was 50%. But then the third flip, it looks like it was tails again because now only one out of three or 33% of the flips have resulted in heads. Now by the fourth flip, we got a heads again, getting us back to 50th percentile. Now the fifth flip, it looks like we got another heads and so now have three out of five or 60% being heads. And so, the general takeaway here is when you have one, two, three, four, five, or six experiments, it's completely plausible that your experimental proportion, your experimental probability diverges from the real probability. And this even continues all the way until we get to our ninth or tenth tosses. But what happens if we do way more tosses. So now I'm gonna do another, well let's just do another 200 tosses and see what happens. So I'm just gonna keep tossing here and you can see, wow look at this, there is a big run getting a lot of heads right over here, and then it looks like there's actually a run of getting a bunch of tails right over here, then a little run of heads, tails, and then another run of heads and notice, even after 215 tosses, our experimental probability has still, is still reasonably different than our theoretical probability. So let's do another 200 and see if we can converge these over time. And what we're seeing in real-time here should be The Law of Large Numbers. As our number of tosses get larger and larger and larger, the probability that these two are very different goes down and down and down. Yes, you will get moments where you could even get 10 heads in a row or even 20 heads in a row, but over time, those will be balanced by the times where you're getting disproportionate number of tails. So I'm just gonna keep going, we're now at almost 800 tosses. And you see now we are converging. Now this is, we're gonna cross 1,000 tosses soon. And you can see that our proportion here is now 51%, it's getting close now, we're at 50.6%. And I could just keep tossing, this is 1100, we're gonna approach 1200 or 1300 flips right over here. But as you can see, as we get many many many more flips, it was actually valuable to see even after 200 flips, that there was a difference in the proportion between what we got from the experiment, and what you would theoretically expect. But as we get to many many more flips, now we're at 1,210, we're getting pretty close to 50% of them turning out heads. But we could keep tossing it more and more and more and what we'll see is, as we get larger and larger and larger, it is likely that we're gonna get closer and closer and closer to 50%. It's not to say that it's impossible that we diverge again, but the likelihood of diverging gets lower and lower and lower the more tosses, the more experiments you make.

1582887065

**Description**

Imagine that every time you speak, people perceive you as a confident and authoritative leader. Wouldn’t it be nice to know that you can speak in a confident manner and in a way that is instantly understandable and memorable to your audience?

In this How to Speak Like a Leader course you will learn the following:

How to avid the most common speaking blunders that undermine authority

How to increase eye contact with your audience

How to structure your presentation so that you don’t seem like a mid-level bureaucrat

How to prepare in the least amount of time possible

This course is delivered primarily through spoken lecture. Because the skill you are learning is speaking related, it only makes sense that you learn through speaking.

The skill you will learn in this class is not primarily theoretical or academic. It is a skill that requires physical habits. That is why you will be asked to take part in numerous exercises where you record yourself speaking on video, and then watching yourself. Learning presentation skills is like learning how to ride a bicycle. You simply have to do it numerous times and work past the wobbling and falling off parts until you get it right.

This course contain numerous video lectures plus several bonus books for your training library.

TJ Walker has been coaching and training people on their presentation skills for 30 years. Now, through the power of Simpliv’s online platform, he is able to give you the same high quality training that he gives in person to CEOs, Fortune 500 executives, and Presidents of countries. Only you can now receive the training at a tiny fraction of the normal fee for in-person training.

How long this course takes is up to you. The longest part of the course involves you speaking on video, critiquing yourself, and doing it over until you like it. But if you get to the point where you love how you look and sound when you present it will be well worth the time spent. And having this skill will save you time for all future presentations in your life.

You can begin improving your leadership presentation skills right now. You may have an opportunity to speak out as soon as tomorrow, so why waste another day worried that your presentation skills are not up to high standards. Enroll in this course today.

There is a 100% Money-Back Guarantee for this course. And the instructor also provides an enhanced guarantee.

What others say:

“TJ Walker’s single-minded devotion to presentation has made him the #1 expert for executives seeking guidance on speaking to the public and media." Bob Bowdon, Anchor/Reporter, Bloomberg Television

“TJ Walker is the leading media trainer in the world." Stu Miller, Viacom News Producer

(TJ Walker’s Media Training Worldwide) “The world’s leading presentation and media training firm."Gregg Jarrett, Fox News Channel Anchor

Who is the target audience?

Anyone how is a leader or who aspires to be a leader

Leaders who wish to speak more effectively

**Basic knowledge**

Students will need to record themselves speaking using a cellphone camera or webcam

**What will you learn**

Speak with the confidence and authority of a leader

Project competence

Present ideas in an understandable manner

Make your key ideas memorable to your audience

#Leadership and Management Online Courses #Learn Leadership with Online Leadership Courses #Leadership Courses #Online Leadership Course