Predict NFL Touchdowns - Create Your First Predictive Model in Python (Step by Step Tutorial)

-------------------------------------------------------------------------------------------------------------------------
Nick Wan is the Director of Analytics for the Cincinnati Reds. He streams data science on Twitch and reacts to the latest news, sports, memes and everything in between.

In this video, Nick teaches you how to make an NFL analytics predictive model that predicts touchdowns. Follow along this walkthrough on Github/Google Collab!

WATCH LIVE ON TWITCH:

Music Licensed by Soundstripe

Video Editor: Tony Pelleriti

[tweakyapollo]

I am a 35 year old intern starting a career in analytics. Thank you for this

[daniellepashayan]

@4:01 - create a list of the seasons you want to include and do a nested list comprehension using any. this helps too so if you ever want to add more seasons, you just edit the variable value instead of changing the code in the list comp
desired_seasons = ['2019', '2020', '2021']
target_seasons = [season for season in seasons if any(target in season for target in desired_seasons )]

[briankkrieger]

I'm a complete noob, but I'm trying to learn. Do you have a video or instructions on how you download the github data so you can use it when following along in google collab

[longieboy]

Very nice! Could you bring more videos on prediction sport outcomes with python?

[Rapha_Carpio]

13:56 maybe you can get rid of all nan values.... just dropping it or fill those NaN with the mean..

[ajjohnson]

I found this tutorial very helpful and really helped a beginner/student like myself. Keep it up!

[HoofSC]

szns = ['2019', '2020', '2021']
target_seasons = [x for x in szn_folders if x in szns]

[WingfanSports]

Great video. For anyone that cares Matt Stafford no only switched teams but also played in the super bowl that year giving him 4 additional games. This model would be significantly better if you limited it to regular season only and maybe strip out players who switched teams

[WingfanSports]

This was Herbert second season so a big jump not surprising. Adding some sort of age curve to this might also be smart

[Rapha_Carpio]

15:02 Previous interceptions has nothing to do with the "next year touchdowns" because, the interceptions might be fault of the offensive player that did not catch the ball... or did not read the play very well and the defend just got the ball on the go...

for newbies... we see that the correlation is so weak in that graphs "touchdowns and interceptions_prev" the line blue is not so high.

[peterk]

Thanks Nick. This was great! Hope that you do more in the future. 🙂

[Rapha_Carpio]

25:42 Can we like put colors to the top 5 Preds so if we are like looking to "buy" a new quarterback we can go for the top 5...
I don't think you need all the 20 seasons, usually a player play at the top level 5-7 seasons..

[colinbehan]

It’s a shame you don’t respond to anyone in here asking questions for support. Not a good strategy for building a following!

[briankkrieger]

Also is there a resource for you to learn what pandas, matplotlib, and seaborn do for you?

[tyler_russell]

If you have more data, why wouldn’t you train with it? For example, why not 2019 and 2020 to train and then test on 2021.

[tweakyapollo]

Does this only work with csv? Or can this be an Excel sheet

[smasher]

I'm having some trouble with the loop and append section is there any other documentation to follow? Using VSCode for reference

[hdie]

Love this type of content, thanks Nick!