Challenge #103: Just another game?

[mithily]

Spoiler

finished this

[gawa]

I end up with just creating linear regression model as I know nothing about football, which is not famous in Japan

[Erin]

Spoiler

My model mirrors the posted solution heavily. However, my goal with this challenge was to gain more experience with data investigation tools. Glad to use the Association Analysis tool and I think I actually understand why the variables were used in the model. 

[Hammad_Rashid]

[RWvanLeeuwen]

revisiting to prep for the expert exam

using R4.2.3 it is really hard to do predictive analytics to due absence of RevoScaleR, weird issues with HTML GUI and models breaking down.... time to manage our own R environments I guess

Spoiler

I love the copy to macro functionality. helps to organise the workflow some more ..allowing the use of all data results in a very poor model fit, so I treat it as just another game

[IgnacioVilaplana]

Attached my solution!

Spoiler

 

[Tgigs]

My solution attached.

Used all regular season games and the following 4 attributes to train the model - opponent score, total yards, passing yards, rushing yards.

Accuracy % on the training data was 80% and accuracy on the testing set (2 superbowl games) was 81%.

Based on that, you would think it is just another game. But it is too small of a sample size to make definitive conclusions. More investigation required.

[Angelica_Obi]

Spoiler

[terry10]

Data types tripped me up at first.