In an off-topic post we would like to share a series of horror narrations based on Win Vector LLC’s very own Nina Zumel’s translations of Uruguayan author Horacio Quiroga. This is a free series produced by Rue Morgue
The first is: “The Feather Pillow.” DO NOT LISTEN TO THIS IN BED!
For quite a while we have been teaching estimating variable re-encodings on the exact same data they are later naively using to train a model on, leads to an undesirable nested model bias. The vtreat package (both the R version and Python version) both incorporate a cross-frame method that allows one to use all the training data both to build learn variable re-encodings and to correctly train a subsequent model (for an example please see our recent PyData LA talk).
The next version of vtreat will warn the user if they have improperly used the same data for both vtreat impact code inference and downstream modeling. So in addition to us warning you not to do this, the package now also checks and warns against this situation. vtreat has had methods for avoiding nested model bias for vary long time, we are now adding new warnings to confirm users are using them.
We had such a positive reception to our last Introduction to Data Science promotion, that we are going to try and make the course available to more people by lowering the base-price to $29.99. We are also creating a 1 month promotional price of $20.99. To get a permanent subscription to the course for less than $21 just visit this link https://www.udemy.com/course/introduction-to-data-science/ and use the discount code ITDS21 any time in January of 2020.
I’d like to share some new timings on a grouped in-place aggregation task. A client of mine was seeing some slow performance, so I decided to time a very simple abstraction of one of the steps of their workflow.