![David Sumpter](/img/default-banner.jpg)
- 28
- 50 693
David Sumpter
Приєднався 12 чер 2018
David Sumpter's UA-cam channel. Follow me on Twitter to keep up to date with Soccermatics, Outnumbered, The Ten Equations and more...
Chris Reid (Macquarie University) presenting his recent work on collective behaviour in weaver ants
Chris Reid (Macquarie University) presenting his recent work on collective behaviour in weaver ants
Переглядів: 113
Відео
Lesson 10: Using Streamlit to Create a Football Analytics Webpage
Переглядів 2,3 тис.Рік тому
This video is part of the Soccermatics course. soccermatics.readthedocs.io/en/latest/
Lesson 7: Evaluating off-ball actions
Переглядів 747Рік тому
We look at how we can combine pitch control with expected threat to evaluate off-ball actions. Video for the lesson soccermatics.readthedocs.io/en/latest/lesson7/ValuingOffBall.html
Lesson 6: Pitch control
Переглядів 1,3 тис.Рік тому
Video for lesson 6 of Soccermatics course. soccermatics.readthedocs.io/en/latest/lesson6/PitchControl.html
Lesson 5a: Binomial and Poisson distributions
Переглядів 707Рік тому
Lesson 5a: Binomial and Poisson distributions
Lesson 4a: Position-based Expected Threat
Переглядів 1,4 тис.Рік тому
Lesson associated with this page: soccermatics.readthedocs.io/en/latest/lesson4/xTPos.html
Lesson 4b: Action-based Expected Threat.
Переглядів 1,1 тис.Рік тому
The video coupled to lesson 4b of Soccermatics: The Course soccermatics.readthedocs.io/en/latest/lesson4/xTAction.html
Lesson 2: Statistical models of football
Переглядів 7 тис.Рік тому
Lesson 2: Statistical models of football
where do you get the data from
Thanks for your sharing, Sorry but I didn’t understand how to finalise this Markov chain when go “lost”
I can do that from scratch using a laptop
00:01:35 «Менеджер должен донести проблему для игроков» Напомнило «Нашу Рашу»: «Уроды! Козлы! Безногие! Всех порешу! Косоногие! Кто ж так играет!». И всё всем понятно... см.сами: ua-cam.com/video/-L2RnETHlT8/v-deo.html
Super cool! When do you expect this to be widely available?
At 27:30, the heatmap for shots has a 'cooler' band along the very top of the penalty area. Is this caused by some kind of data error/edge-case, or do players really avoid shooting from on the painted line?!
awesome, congrats! :-)
brilliant
Hi David, I am a full-time student of Statistics. In my spare time, I read your book "Soccermatics" it is absolutely amazing and inspiring to become a Soccer analyst. Could you please upload a full playlist on "Soccer analysis using R" or "How to become a Soccer analyst", also, a course or a playlist on the required concepts of Mathematics and Statistics to become a sports analyst using R programming language.?
Hey I'm a math major. What courses do you recommend to become soccer analyst
😮 This one is quite different from the on Wikipedia. Sir, Did you create this yourself? As a student of statistics, I am currently reading your book "Soccermatics" during my spare time. It is amazing 😍. Regards: Sanaullah Khan, from Pakistan.
Nice tool! I love the whole philosophy behind Soccermatics. Thank you very much!
Hi Sir ! I've learnt a lot from you this past year , In my country there is not such any type of education of Football Analytics in my country which is India. So, your content have been a great source for me to learn about this field and not only I'm learning , I'm also putting out my work Indian Football Fanatics to make them aware about this field using my knowledge. I'm one of the fewest who is doing this on youtube. Just want to say thanks for your time and efforts you have put to aware more people !!
Great explanation🎉🎉🎉
Thank you !
Thanks so much for this video and the explanation, a lot of the visualizations wouldn't make full sense without the context and this video was a great provider of that too.
Hello David, thank you for the informative videos. Do you by chance have a dataset of expected threat to work with? I can't find xThreat data anywhere online. Thanks
This is really great, thank you David! Really grateful for the Soccermatics course.
Amazing
Brilliant. Thank you so much for sharing your knowledge.
this is just an amazing work continue on these valuable lectures
this models do not account for the randomness of human behaviour. as a player/team becomes more well drilled and skilled they will begin to create deviation in these models and de. how about assigning different points/weights for each player to account for these randomness.
So I might be off by a mile, but are top teams and our dear Prof. Sumpter feeding this to ML to get models and results? Or is this more of a simulation of current data?
This is quite remarkable. I was aware that pro teams have data analyst teams and models that capture movement. But also positional effect? And, with domain knowledge, also optimal tactics? This is breathtaking on my first sight.
I am very sorry if I am coming off as too forward. I read your literature for The Athletic, and I am super excited to explore this further. Please let me know if we could connect. I promise I won't take much of your time, I promise.
Greetings Professor. Now that I have finished all your videos, allow me to request if we could perhaps discuss modelling threat evaluation. I am privileged to learn so much from you, and I have an idea I have been writing on for a bit. I would be happy to keep it on the UA-cam comments, but is there some perhaps email we could connect on? May I leave my email where we could connect?
🇮🇳
Such clear explanations in this video. Probability and algebra are a formidable force in mathematics.
Absolutely awesome!!!
I wanted to learn the maths required for this. Where can i learn them.
This video
Thank you for the brilliant lessons
Top top content professor 👌👌👌
Firstly, thank you for your efforts. But after trying to make the only forward passes network, l couldn't do it, and I don't know where to get help after searching for a while. It would be nice if you can help me with that. Thank you.
I used lambda to try and find whether a pass was made forwards or not. I took the original dataframe (df) and created a new column based on the following condition: df['pass_direction'] = df[['x', 'end_x']].apply(lambda i: 'forward' if i['end_x'] > i['x'] else 'backward', axis=1) I am making an assumption that if the 'end_x' is greater than the 'x' position for that particular id then it must be a forward pass? Let me know what you think
@@aashiksingh7704 Thank you for your respond, I think that is really a good idea, I already knew the suedo code but I'm not comfortable with data science and python, I appreciate your help, thank you. 😀
@@mohamedkhaled4888 Hopefully you completed the challenge - I changed the code defining the mask. I made the same assumption as Aashik about a foward pass (end_x > x) and this seems to be correct. My solution, I added the last & and the condition following it: mask_england_fwd = (df.type_name == 'Pass') & (df.team_name == "England Women's") & (df.index < sub) & (df.outcome_name.isnull()) & (df.sub_type_name != "Throw-in") & (df.end_x > df.x)
Thanks .Professor David
It was very disappointing by seeing this course is free only for EU CITIZENS 😢.
Thanks from Brazil!
Fun to see that it's not just me who loves football! Keep having fun 👍🏾⚽️ I’m Elie Prince