News from olavla



















I made a sword out of a 2x4

I'm in this with you.

Shows the How did he do that? Award and grants %{coin_symbol}100 Coins to the community. Exclusive to this community.

Shows the Silver Award... and that's it.








  1. If you use separate partitions to fit the individual components and the stack, you shouldn't. Obviously, you get a more expressive model (lower bias) at the cost of higher variance.

  2. In terms of deaths and wounded, according to this data, the trend is declining!

  3. Life pro tip: you are pretty enough to not use this thick of a layer of makeup.

  4. Up to the modeling, the article is nice.

  5. As a woman, you can get away with that, but...you need to have a model like appearance.

  6. That might work for like 6 months. Men won't remain mesmerized by a model like appearance incredibly long. I've dated models and broken up with them because they were sketch with obvious emotional issues.

  7. Some celebs manage to get married and walk with half...

  8. Crossvalidation is an approach to model selection, not perse to prevent overfitting in a single model. As the final model averages the models built on the different subsets, you get some smoothing.

  9. I noticed my r squared was at 99.99. Also my plot was giving very weird results, it was very hard to interpret my actual and predicted values. So I assumed by the minimal knowledge that i have that my model might be overfitted. But reading comments now, i feel that the reason why i got such results was probably because my distribution wasn't supposed to work on linear regression in the first place. But the close proximity between actual and predicted values are confusing me.

  10. Yeah, R2 at that value is correct reason for concern. One of your predictors just correlates 100% with your target. Simply create a scatterplot for your target against each predictor and you will find out.

  11. Option 1) fit a single tree using RF and display that tree

  12. Musk considered this, but it was infeasible for him at the time as the batteries are part of the structural design of the car. Some say that exactly this feature will make the Chinese manufacturers overtake Tesla in the future.

  13. Have your team members create presentations/sharing sessions on topics of your interest.

  14. Would he chuckle if he saw it?

  15. Nothing new or unknown. Greeks have been doing this since 2000 years and the wine that tastes like resin ('retsina') is still sold today.

  16. Put his head on it -> Fool's gold 🤣

  17. Now imagine it was a 45 yo dude and a 14 yo girl. You think they would have used the same wording and images?

  18. Use paragraphs and run! This will never be a happy situation for you.

  19. Proud boys..sounds so much like a organization that helps gays come out.

  20. Just show your garage, nothing interesting about a high school woodwork project though.

  21. The best jobs welcome you as family.

  22. You can use soapy water to determine the source of the leak. Most likely your gas line. Tank will be fine. Check if the gas line has a thread at the other side that came loose.

  23. You can hear a short hiss (max 1 second) that then stops. This is the filling of the regulator and the gas line.

  24. This is just to prevent abortion clinics to be setup on tribal land.

  25. Are you a drugs user, though?

  26. The harder the concepts come to you, the better you will be able to explain them to others due to your own struggles to grasp them. I've met math geniuses that write the most difficult models out without any issue, but fail to understand what it really means, let alone, convey information on them.

  27. Covariance matrix can become non-positive definite when you have multicollinearity. Quick and dirty solution is to use PCA to compress your variables.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may have missed