Feature engineering concepts - Amazon Web Services (AWS) Tutorial

Here we have a traditional feature engineering problem. In this case, I have the NBA box score and it has a column with total wins for the season. One of the things we can do is repurpose this old feature and convert it into a new feature. In this case, we could look at the total wins and determine if it's over a threshold. So depending on how many games there are in a season, you could say, for example, if you have 50 wins, you would have a winning season and that would be a new value called one. If you had 35 wins, depending on how many games are in the season, again, this could be a zero that would go into this new column called winning and losing record. Once you have this new feature, now what you can do is use it for a prediction. So we could use that as the label, right that we're trying to predict. And we would say, for example, can we use how many points a team would average and use that to predict whether they would have a winning record or losing record in the season? Let's…