This blog post contains spoilers for episode 4 of season 8 of RuPaul’s Drag Race, so beware before reading on!
On this week’s episode of Drag Race, the queens are split into three groups and tasked with performing as 80s New Wave bands to a live audience. Robbie Turner, Kim Chi, and Naomi Smalls impress with their food-themed punk band Les Chicken Wings. Robbie Turner is awarded this week’s win for her high energy performance.
Bob the Drag Queen, Acid Betty, and Thorgy Thor are declared safe for their party band performance as Street Meatz. Derrick Barry, Chi Chi DeVayne, and Naysha Lopez landed in the bottom with their lackluster synth band, Dragometry. Derrick was ultimately declared safe, leaving Naysha and Chi Chi to lipsync for their lives to Blondie’s Call Me (a tribute to this week’s guest judges Debbie Harry and Chris Stein). In the end, Naysha had to sashay away for a second, and likely final time.
Last week, the algorithm contestants predicted Naysha would be going home next, and they were correct! Before I get to their new predictions for next week, however…
We’ve got a new contestant joining us! Say hello to Neural Network. Neural networks are a family of methods that roughly simulate connections between neurons in a biological brain. The neural network used here consists of neurons that take some number of values as inputs, applies weights to these values (that can be adjusted in the learning process), then applies these values to a logistic function to produce an output between 0 and 1. Neural networks consist of two or more layers of neurons (an input layer, an output layer, and zero or more hidden layers). The output layer must have the same number of neurons as the number of output values, but other layers can have any number of neurons. The network I use here has two layers, with five neurons in the input layer and fourteen neurons in the output layer (one for each possible place a queen could end up in).
Neural networks perform best if the input values are scaled similarly, so for all but the dummy variables, I scale the values to mean of 0 and standard deviation of 1. As I did with the other algorithms, the neural network is trained with seasons 1 through 6, and tested with season 7. The output of the neural network consists of fourteen values for each queen, each value corresponding to the best guess of the network for whether the queen took that place. To get the final predictions of the network, I grab which place had the highest value and assign that as the predicted placement for the queen. I then rank the ranks to ensure a first place is predicted. So how did Neural Network do?
|Jaidynn Diore Fierce||8||6|
|Mrs. Kasha Davis||11||11|
The table above shows the predicted placement for queens in season 7, and it does extremely well! It switches Trixie and Jaidynn, and mixes up the bottom 3, but otherwise correctly guesses everyone else. It has a Kendall’s Tau of 0.884, doing better at predicting season 7 then Random Forest Regressor.
So, I add Neural Network to the predictions table, computing a weighted average using Kendall’s Tau scores from season 7, and come up with the following predictions:
|Actual||Support Vector Machines||Gaussian Naive Bayes||Random Forest Classifier||Random Forest Regressor||Neural Network||Average Predicted Score|
|Chi Chi DeVayne||6||4||4||2||3||3.355|
|Bob the Drag Queen||6||6||4||2||3||3.680|
|Cynthia Lee Fontaine||10||6||4||12||12||6||8.461|
Naomi Smalls has the lowest average score of the queens remaining, and so is predicted to go home next. She is also at least tied for lowest in all five algorithms, so there is consensus here. Naomi is a queen known for her look and runway, and next week is the Snatch Game, a challenge that requires comedy and wit to succeed, so it would not be a surprise if Naomi has difficulty with it.
The new predicted top three are Kim Chi, Chi Chi DeVayne, and Bob the Drag Queen. Acid Betty has slipped out of the top three thanks largely to Neural Network‘s lower ranking of her.
For further reading: