Probability results from Week 9
| Baker | Winner | Finalist | 3rd-4th Place |
|---|---|---|---|
| David | 29% | 69% | 22% |
| Steph | 25% | 97% | 2% |
| Alice | 22% | 98% | 1% |
DeepBake is a set of deep learning neural network models to predict the final rankings of GBBO contestants.
DeepBake consists of 10 models for each episode, and was trained on data from seasons 2-9.
Data include 8 variables:
- Technical Challenge Ranking for that week and running mean from prior weeks
- Contestant was Star Baker and running mean of times named Star Baker
- Contestant was a favorite baker that week and running mean from prior weeks
- Contestant was an unfavored baker that week and the running mean
Data were obtained from Wikipedia. Thanks to those who made those pages.
Data were then quantile scaled to fit a normal distribution.
DeepBake's performance was measured using a Leave One Out method. One season was set aside for evaluation while training the model on using the remaining seasons. A mean receiver operating curve was calculated by iterating through all seasons.
The closer the area under the curve (AUC) is to 1, the more accurate the model.
Random chance of making a correct prediction has an AUC of 0.5 (dotted diagonal line). The Episode 4 model has an AUC of 0.91 (+/- 0.04 95% Confidence Interval), meaning it has a very good chance of predicting the final GBBO winner!
DeepBake makes 5 predictions:
- 1st Place 🏆 🏆
- Runner-Up 🏆
- 3rd-4th Place
- 5th-7th Place
- 8th Place and Below
The evaluation was measured using this tiered class system.
Note how the classifier gets better at predicting as the season progresses. This makes sense because the good bakers rise to the top (favored and star bakers) and historical data are recorded as running means.
Absolutely! Here are the current standings!
SEMIFINALS and DeepBake did well this time, correctly predicting Rosie woud get the boot with an 80% chance of placing in 3rd-4th place.
So the question remains? Who is going to win GBBO?
Here's a table of the current standings. Let's consider all the probabilities listed here.
| Baker | Winner | Finalist | 3rd-4th Place |
|---|---|---|---|
| David | 29% | 69% | 22% |
| Steph | 25% | 97% | 2% |
| Alice | 22% | 98% | 1% |
David has the highest chance of winning GBBO, but he also given a much higher score for 3rd-4th place than the other two bakers. He is also given a much lower score for being a Finalist than the other two bakers. Steph has been solid for the entire serires but has recently declined in standings. I think Steph is the safe bet while Alice is the Dark Horse.
Alice's chance to win has bounced around quite a bit, which makes her less of a safe bet. But she has come on top in many weeks.
So do the judges really consider all previous bakes too, as they claim? If they do I think it's clear David or Steph will win GBBO. David has a real chance of winning since he has caught up to the juggernaut that is Steph. However, David has also fudged the rules sometimes (I mean a pie without a cover??? Really mate?). Steph has won Star Baker 4 times while David has never been graced by that honor. There has never been a GBBO winner that has not won Star Baker. Just let that bake for a minute.
Pastry week did not go well for the bakers, nearly everyone had some problems in the signature or in the technical (cough Steph) or at the showstopper. What does this mean for David, who rocked the obscure Moroccan technical? Well now he is in the lead for winning GBBO at a 22% chance, while Steph dropped down to 21%. Still neck-and-neck for those two.
As for the rest of the pack, Henry and Alice swapped places again: now Alice is poised to be a GBBO Finalist (75%) as opposed to Henry's 75% chance he got at week 7.
In fact, Henry left the tent this week and DeepBake did not predict it. Giving Rosie the highest chance of placing 3rd through 4th place with 89% probability. Henry was close to follow with a 70% chance. So in defense of DeepBake, it has correctly predicted who will leave given the highest or second highest odds to leave for every week thus far.
Stay tuned for next week, the semi-final, and see if DeepBake has correctly pegged who will be a GBBO finalist. Will it be Steph, David, and Alice, or will DeepBake
If you wanna be a GBBO winner, you gotta make consistent bakes. And that's something Alice needs to work on if she wants to make it to the final. She did not do well in the Signature and the Technical, plummeting her odds to win GBBO. As the wheel of fortune turns downward for Alice, it raised Henry up from a 20% chance last week to 75% chance this week to bake in the final.
As for predicting who will leave the tent, DeepBake missed this one. It thought Rosie would leave with a 66% chance, but Michael was the unlucky baker who was eliminated. His chance to leave was pegged at 43%, second to leave according to DeepBake. Now DeepBake has predicted 6 out of the 8 bakers who would leave, which is a 75% chance of getting this prediction correct. This is not so bad, but hopefully DeepBake can do better, even though Micheal was predicted next to leave after Rosie.
Steph is still solidly leading the GBBO winner odds with 40% chance to win, following her is David with 35% chance and then Henry with a 24% chance. It seems the spread between Steph and David is closing in as the bakers inch toward the final.
What I find intriguing is how the 3rd-4th place category is not lateral compared to last week. Now Alice is leading this group with 34% chance to place in either 3rd or 4th place. This means DeepBake thinks she will not get eliminated next week, but will not bake in the final. Henry and Alice (who are lovebirds by the way), are not consistent bakers which makes their chance of winning less believable than Steph or David. However, I think we can all agree that Rosie will likely leave the tent next week.
Well lads, it happened, Priya has left the tent. DeepBake gave her a whopping 81% to place in 5th-7th place, Rosie is next with 65% and Henry with a 59% chance.
Alice had a great week 6, winning the technical and being favored by the judges. Her chances to win GBBO has rocketed to 35% from 8% last week. She is behind Steph (42%) who has solidly secured star baker for three weeks in a row.
DeepBake's predictions for 3rd and 4th place are lateral for the bakers, but perhaps this is expected since evident by the first three weeks of baking for 5th to 7th place.
DeepBake seems to be on a roll (or a bap), it has predicted who will leave 6 out of 7 times which is 86% accuracy close to the 90% estimate I made on reddit.
First off, I realize I have been using "finalist" incorrectly, now I have labeled plots as "GBBO Winner" and "GBBO Finalist". The finalist probability is the chance a baker will compete in the final week.
This week my wife and I paused the show right after Steph was announced star baker and ran the model with the new data. Steph has been starbaker twice and been favored by the judges for two other weeks. Therefore, her standings as a GBBO finalist and series winner is gaining ground with a 46% and 77% chance respectively.
For the outright winner, Henry is next in line with a 28% chance and then is David with 17%. Likewise, David is next in line to be a finalist with 70% probability. Then comes Michael (60%). Last week, DeepBake thought Michael would be next on the chopping block with Henry, but both bakers are doing well this week. When it comes to predicting who will leave next week, DeepBake made some misses but also some hits.
As I mentioned, my wife and I entered the new data into the model after Steph was crowned star baker to predict who would leave. This week it was obvious two bakers had to go, so looking at the model Priya was heads above all with 27% chance of leaving. Following her was Helena with 15% chance to leave. The chances for the remaining bakers quickly decay with probabilities less than 3%.
Priya was not eliminated, but you have to admit she probably should have been. Helena was eliminated and so was Michelle. If we look at the 5th to 7th place predictions, which we should since we are at the point of the series to use that score to predict who might leave the tent, we see Michelle and Helena with 71% and 63% probability respectively. Next up is Pryia with a 55% chance to leave; Rosie and Alice tail her with 36% and 33% chances.
What can I say, even Neural Networks make mistakes. The top baker prediction now thinks Rosie is in the lead with 23% chance, with Alice following behind her with a 18% chance. Something about Alice is really making the model put her at the top, which I don't really agree with.
Anyhoo, the finalist (top 3 bakers) predictions seem to make more sense with Steph (75%), David (64%), and Rosie (43%). Alice is still up there and so is Michael.
As for the bottom bakers, last week the model said Henry had a good chance of leaving and he did have some pretty hard bakes in week 4. However, Phil was eliminated and the week 3 model did not put him anywhere near the bottom... So DeepBake got it wrong in week 3. But the week 4 predictions put him right at the bottom, which means if I were to pause the show right after the judges reveal who is star baker and input all the data needed for DeepBake, the model would have made a correct prediction that Phil would leave.
Henry is slated to leave again with a 30% chance, not sure why Henry is not favored by the model too. Oddly, Michael is right behind him, probably because he got an unfavorable status from the judges in week 4. Priya is next and so is Michelle, which makes sense to me.
DeepBake now thinks Alice hasn't proven herself and her standing has slid to danger territory (week 2). David is now the most likely winner and Steph jumped up in the rankings.
As for the finalists (top 3 bakers), Michael has a 67% chance of baking in the final, DeepBake thinks Steph has a 59% chance and David a 47% chance.
DeepBake in week 2 gave Amelia a 48% of being in the bottom tier, right behind the two eliminated bakers Dan and Jamie. In week 3 Amelia was eliminated, which is bad news for her but great news for DeepBake!
For remaining bakers in week 3, Amelia was at the top of the ranking for the bottom tier with a 43% probabiliy. Next is Henry, (winner of the bap technical) with a 36% probability. Following him is a tight pack led by Priya with Helena, Michelle, and Alice following her, all with probability scores between 23-20%.
DeepBake puts Alice in the lead with a 36.8% probability score for being the finalist. Michael, David, and Rosie are close contenders with around 21% probability.
This score is the addition of the finalist probability and the runner-up probability. It's a measurement of how likely a baker would be in the final episode.
DeepBake thinks Alice (87%), David (62%), and Michael (60%) will vie for the title of best baker.
DeepBake gave Dan and Jamie the highest scores (80% and 78%) for being in the bottom tier. Dan was eliminated in week 1, while Jamie was eliminated at the end of episode 2.
In fact, DeepBake can make a prediction before the judges eliminate a baker. These results suggest DeepBake correctly predicted Jamie would leave the tent!