Translate
Tuesday, November 17, 2020
Best Black Friday 2020 headphone sales: Samsung, Sony, Beats, Jaybird and more deals available now - CNET
from CNET News https://ift.tt/38MeApO
via A.I .Kung Fu
Home Depot Black Friday sale is on: Ryobi drill set for $199, Honeywell smart thermostat for $99 and more - CNET
from CNET News https://ift.tt/3kzKZSG
via A.I .Kung Fu
Black Friday Apple Watch deals bring lowest prices ever: Get $49 off Apple Watch SE and Series 6 now, Series 3 drops to $119 starting Nov. 25 - CNET
from CNET News https://ift.tt/36GDogt
via A.I .Kung Fu
Black Friday 2020 TV deals: LG and Sony OLED TVs on sale, $398 Samsung 58-inch available now, $478 Vizio 70-inch coming soon - CNET
from CNET News https://ift.tt/2Uv2TM2
via A.I .Kung Fu
Newegg Black Friday deals: Discounts smart TVs, gaming PCs, laptops and much more - CNET
from CNET News https://ift.tt/3kJr304
via A.I .Kung Fu
NASA spots wild dust devil tracks on Mars that look like claw marks - CNET
from CNET News https://ift.tt/36GhB8y
via A.I .Kung Fu
Best Buy Black Friday deals available now: $530 70-inch Samsung 4K TV, $160 Powerbeats headphones and more to come - CNET
from CNET News https://ift.tt/35wEYlu
via A.I .Kung Fu
Trump fires the US's leading authority on election security - CNET
from CNET News https://ift.tt/32PMnuw
via A.I .Kung Fu
Lenovo Black Friday sale: Save on ThinkPad laptops and more - CNET
from CNET News https://ift.tt/2IKsj5q
via A.I .Kung Fu
12 best TV shows to binge-watch on Disney Plus - CNET
from CNET News https://ift.tt/3lGplNX
via A.I .Kung Fu
From Apple to Samsung: 5G phones available right now - CNET
from CNET News https://ift.tt/35FF6PV
via A.I .Kung Fu
The 17 best TV shows to binge-watch on Amazon Prime Video - CNET
from CNET News https://ift.tt/3lF7KGs
via A.I .Kung Fu
Target Black Friday 2020 ad scans: Save on Apple Watch, games and more - CNET
from CNET News https://ift.tt/32NsLas
via A.I .Kung Fu
Gamer born without a hand gets Metal Gear Solid Venom Snake bionic arm - CNET
from CNET News https://ift.tt/35Dv4P1
via A.I .Kung Fu
New watch contains pieces of Stephen Hawking's desk - CNET
from CNET News https://ift.tt/2H99ad4
via A.I .Kung Fu
The Crown season 4: All your questions answered and that ending explained - CNET
from CNET News https://ift.tt/3nB8M70
via A.I .Kung Fu
Best PC speakers for 2020 - CNET
from CNET News https://ift.tt/3glxTXq
via A.I .Kung Fu
5G phones in 2020: iPhone 12, Galaxy Note 20, Pixel 5 and more - CNET
from CNET News https://ift.tt/36IEnwt
via A.I .Kung Fu
The Queen's Gambit: That ending explained and all your questions answered - CNET
from CNET News https://ift.tt/354MK5P
via A.I .Kung Fu
The Queen's Gambit could be a game changer for women's chess - CNET
from CNET News https://ift.tt/2IMeqEf
via A.I .Kung Fu
Tweets making fun of Twitter's new Fleets will not vanish in 24 hours - CNET
from CNET News https://ift.tt/2UBrQoZ
via A.I .Kung Fu
2021 Mercedes-AMG GT Black Series is your new king of the Nürburgring - Roadshow
from CNET News https://ift.tt/38S3qQD
via A.I .Kung Fu
What's Section 230? Everything you need to know about free speech on social media - CNET
from CNET News https://ift.tt/2JjDliD
via A.I .Kung Fu
Google adds more COVID-19 details to its Maps app as cases surge - CNET
from CNET News https://ift.tt/3nBc0qL
via A.I .Kung Fu
Twitter CEO Jack Dorsey to urge lawmakers to build on key internet law - CNET
from CNET News https://ift.tt/36DfQsM
via A.I .Kung Fu
The best battery-powered home security cameras to buy this year - CNET
from CNET News https://ift.tt/3f8ckui
via A.I .Kung Fu
Monday, November 16, 2020
Huawei is selling off Honor phone business to 'ensure its own survival' - CNET
from CNET News https://ift.tt/3kLFFfr
via A.I .Kung Fu
The Crown season 4: What it gets right (and wrong) about Princess Diana - CNET
from CNET News https://ift.tt/2KfKOjb
via A.I .Kung Fu
The Crown season 4: What it gets right (and wrong) about Margaret Thatcher - CNET
from CNET News https://ift.tt/35BeShb
via A.I .Kung Fu
UFC 255: Figueiredo v Perez -- Start time, how to watch online and full fight card - CNET
from CNET News https://ift.tt/38LQQ5j
via A.I .Kung Fu
2021 Jeep Wrangler Rubicon 392 is a V8 victory - Roadshow
from CNET News https://ift.tt/2KfKJMp
via A.I .Kung Fu
Facebook labels reportedly ineffective at confining Trump's false election claims - CNET
from CNET News https://ift.tt/3f7qBY4
via A.I .Kung Fu
The Crown season 4 ending explained, and all your questions answered - CNET
from CNET News https://ift.tt/2IGsP4K
via A.I .Kung Fu
Black Friday deals on smart home devices: $199 Shark Robot Vacuum available now, $19 Nest Mini coming next week - CNET
from CNET News https://ift.tt/3lJvidm
via A.I .Kung Fu
Best gifts for moms in 2020 - CNET
from CNET News https://ift.tt/2Kj7Iq1
via A.I .Kung Fu
Best cash-back credit cards for November 2020 - CNET
from CNET News https://ift.tt/3f3RMmN
via A.I .Kung Fu
Black Friday 2020 TV deals: LG and Sony OLED TVs on sale, $398 Samsung 58-inch available now, $478 Vizio 70-inch coming soon - CNET
from CNET News https://ift.tt/2Uv2TM2
via A.I .Kung Fu
Best cases for iPhone 12 and iPhone 12 Pro - CNET
from CNET News https://ift.tt/3kjnXAl
via A.I .Kung Fu
Barack Obama shares a playlist of favorite songs from White House years - CNET
from CNET News https://ift.tt/3lEknBp
via A.I .Kung Fu
The best face masks for running outside - CNET
from CNET News https://ift.tt/2HbRVYO
via A.I .Kung Fu
The Crown: Here's how to spot that real mouse darting through a scene - CNET
from CNET News https://ift.tt/3pD9ZfB
via A.I .Kung Fu
Airbnb files for IPO, shows it can actually make a profit - CNET
from CNET News https://ift.tt/3nzV9of
via A.I .Kung Fu
Xbox Game Pass: 18 awesome Xbox and PC games to play right now - CNET
from CNET News https://ift.tt/2H7XBTv
via A.I .Kung Fu
MagSafe on iPhone 12: I still want USB-C, but I was wrong about Apple's magnetic charger - CNET
from CNET News https://ift.tt/2K8RBuR
via A.I .Kung Fu
Sunday, November 15, 2020
Pininfarina Battista is smarter than your phone when roaming - Roadshow
from CNET News https://ift.tt/3kAkTPN
via A.I .Kung Fu
The best advent calendars that include food and drinks for 2020 - CNET
from CNET News https://ift.tt/2Uxa9qw
via A.I .Kung Fu
Newegg Black Friday deals: Save huge on an 85-inch Samsung TV, gaming PCs, laptops and more - CNET
from CNET News https://ift.tt/35zVxx0
via A.I .Kung Fu
SpaceX, NASA Crew-1 mission makes historic launch to the ISS - CNET
from CNET News https://ift.tt/38LXZm0
via A.I .Kung Fu
Best student credit cards for November 2020 - CNET
from CNET News https://ift.tt/2II5m3c
via A.I .Kung Fu
Best stocking stuffer ideas: Gifts under $25 - CNET
from CNET News https://ift.tt/2UxtOqx
via A.I .Kung Fu
Best Black Friday 2020 laptop deals: Huge Savings on HP, Lenovo, Microsoft Surface and more - CNET
from CNET News https://ift.tt/3nskcK5
via A.I .Kung Fu
Amazon Echo early Black Friday deals: $75 Show 5, $13 Echo Flex and more - CNET
from CNET News https://ift.tt/2K42XjK
via A.I .Kung Fu
The 30 best iPad games you need to play - CNET
from CNET News https://ift.tt/38JKeUS
via A.I .Kung Fu
PS5 launch games: All the PlayStation 5 titles you can buy now - CNET
from CNET News https://ift.tt/3pssP9r
via A.I .Kung Fu
Home Depot Black Friday sale is on now: See all the best deals on refrigerators, drills, and more - CNET
from CNET News https://ift.tt/3kzKZSG
via A.I .Kung Fu
Avatar 2: Kate Winslet is 'very proud' of breaking Tom Cruise's underwater record - CNET
from CNET News https://ift.tt/2UutMj5
via A.I .Kung Fu
The Queen's Gambit: That ending explained and all your questions answered - CNET
from CNET News https://ift.tt/354MK5P
via A.I .Kung Fu
Best Black Friday AirPods deals: Airpods Pro at $200, a $49 savings - CNET
from CNET News https://ift.tt/3lN4r01
via A.I .Kung Fu
8 PS5 UI tips and tricks to get the most out of your new PlayStation - CNET
from CNET News https://ift.tt/32Nkm6M
via A.I .Kung Fu
Best Buy Black Friday 2020 ad: Huge sales on TVs, Speakers, Chromebooks, and more all month long - CNET
from CNET News https://ift.tt/3nomZ75
via A.I .Kung Fu
17 of the best TV shows to stream on Amazon Prime Video - CNET
from CNET News https://ift.tt/2IvWAFL
via A.I .Kung Fu
PS5 vs Xbox Series X: The consoles we're buying and in which order - CNET
from CNET News https://ift.tt/32MymO5
via A.I .Kung Fu
The Crown season 4 ending explained, and all your questions answered - CNET
from CNET News https://ift.tt/3noVZnS
via A.I .Kung Fu
The 12 best TV shows to binge-watch on Disney Plus - CNET
from CNET News https://ift.tt/3pycXSv
via A.I .Kung Fu
The 32 best movies to stream on Disney Plus - CNET
from CNET News https://ift.tt/36DJpdJ
via A.I .Kung Fu
31 of the best TV shows to see on Hulu - CNET
from CNET News https://ift.tt/2H4EXvT
via A.I .Kung Fu
OnePlus 9 may sport a bigger screen, triple camera setup - CNET
from CNET News https://ift.tt/3kzf4lF
via A.I .Kung Fu
Best robot vacuums for 2020: iRobot Roomba, Eufy, Electrolux, Neato and more - CNET
from CNET News https://ift.tt/2Uraw6d
via A.I .Kung Fu
Whole Foods adds a new plant-based bacon - CNET
from CNET News https://ift.tt/3kA473d
via A.I .Kung Fu
NFL 2020: How to watch Seahawks vs. Rams, Bills vs. Cardinals, RedZone and the rest of Week 10 today without cable - CNET
from CNET News https://ift.tt/36CppZ0
via A.I .Kung Fu
The incredible Pocketalk Classic mobile translator is on sale for $99 - CNET
from CNET News https://ift.tt/3khy47R
via A.I .Kung Fu
Epic's free game this week is a 'bullet hell' action game with a lot of typing - CNET
from CNET News https://ift.tt/3ndmyfE
via A.I .Kung Fu
Apple Watch Series 6 vs. Fitbit Sense: Top smartwatches go head to head - CNET
from CNET News https://ift.tt/35voBpi
via A.I .Kung Fu
Apple HomePod Mini review: iPhone users will love this $99 Siri smart speaker - CNET
from CNET News https://ift.tt/38KvzJd
via A.I .Kung Fu
Apple's remaking Mac computers, and it's taking control to do it - CNET
from CNET News https://ift.tt/3lzM6TN
via A.I .Kung Fu
OnePlus Nord N10 5G review: An affordable 5G phone with few compromises - CNET
from CNET News https://ift.tt/2UqAp69
via A.I .Kung Fu
Best car plastic restorer: Chemical Guys, Meguiar's and more compared - Roadshow
from CNET News https://ift.tt/3f8cmlR
via A.I .Kung Fu
Facebook, Twitter CEOs to testify before the Senate: How to watch Tuesday - CNET
from CNET News https://ift.tt/3lyhBxN
via A.I .Kung Fu
MacOS Big Sur: Apple's new M1 chip will make apps run faster and smoother - CNET
from CNET News https://ift.tt/3nqTVMi
via A.I .Kung Fu
Pixel 5 specs vs. iPhone 11, Galaxy S20 FE and OnePlus 8 - CNET
from CNET News https://ift.tt/3cIve9Z
via A.I .Kung Fu
Best cheap gaming laptops under $1,000 for 2020 - CNET
from CNET News https://ift.tt/3jTbhze
via A.I .Kung Fu
Best laptop for 2020 - CNET
from CNET News https://ift.tt/2D6AxTc
via A.I .Kung Fu
Google Pixel 5's wimpy camera is driving me to the iPhone 12 - CNET
from CNET News https://ift.tt/3kxy2cm
via A.I .Kung Fu
iPhone 12 Mini review: Apple's smallest is a one-handed phone user's dream - CNET
from CNET News https://ift.tt/36Ax75L
via A.I .Kung Fu
iPhone 12 vs. Pixel 5: Apple and Google's 5G flagships compared - CNET
from CNET News https://ift.tt/2UsRKv1
via A.I .Kung Fu
Saturday, November 14, 2020
The best Prime Day 2020 deals still available: Get a MacBook Air for $850, a Roku for $27, AirPods Pro for $199 - CNET
from CNET News https://ift.tt/3nXq6DX
via A.I .Kung Fu
Qualcomm gets OK to sell 4G chips to Huawei, despite US ban, report says - CNET
from CNET News https://ift.tt/3ktfQAB
via A.I .Kung Fu
The best gifts for girls ages 9-12 - CNET
from CNET News https://ift.tt/3mHBiTO
via A.I .Kung Fu
Friday, November 13, 2020
Affordable holiday gift guide for car lovers in 2020 - Roadshow
from CNET News https://ift.tt/2IDqrvR
via A.I .Kung Fu
Solv Health, which sells patient management software to health care providers, mainly urgent care clinics, raises $27M Series B+ led by Acrew Capital (Kia Kokalitcheva/Axios)
Kia Kokalitcheva / Axios:
Solv Health, which sells patient management software to health care providers, mainly urgent care clinics, raises $27M Series B+ led by Acrew Capital — Solv Health, a startup that sells health care providers digital tools to manage patients, has raised $27 million in new funding led by Acrew Capital …
from Techmeme https://ift.tt/3pDa4jE
via A.I .Kung Fu
Predicting qualification ranking based on practice session performance for Formula 1 Grand Prix
If you’re a Formula 1 (F1) fan, have you ever wondered why F1 teams have very different performances between qualifying and practice sessions? Why do they have multiple practice sessions in the first place? Can practice session results actually tell something about the upcoming qualifying race? In this post, we answer these questions and more. We show you how we can predict qualifying results based on practice session performances by harnessing the power of data and machine learning (ML). These predictions are being integrated into the new “Qualifying Pace” insight for each F1 Grand Prix (GP). This work is part of the continuous collaboration between F1 and the Amazon ML Solutions Lab to generate new F1 Insights powered by AWS.
Each F1 GP consists of several stages. The event starts with three practice sessions (P1, P2, and P3), followed by a qualifying (Q) session, and then the final race. Teams approach practice and qualifying sessions differently because these sessions serve different purposes. The practice sessions are the teams’ opportunities to test out strategies and tire compounds to gather critical data in preparation for the final race. They observe the car’s performance with different strategies and tire compounds, and use this to determine their overall race strategy.
In contrast, qualifying sessions determine the starting position of each driver on race day. Teams focus solely on obtaining the fastest lap time. Because of this shift in tactics, Friday and Saturday practice session results often fail to accurately predict the qualifying order.
In this post, we introduce deterministic and probabilistic methods to model the time difference between the fastest lap time in practice sessions and the qualifying session (∆t = tq-tp). The goal is to more accurately predict the upcoming qualifying standings based on the practice sessions.
Error sources of ∆t
The delta of the fastest lap time between practice and qualifying sessions (∆t) comes primarily from variations in fuel level and tire grip.
A higher fuel level adds weight to the car and reduces the speed of the car. For practice sessions, teams vary the fuel level as they please. For the second practice session (P2), it’s common to begin with a low fuel level and run with more fuel in the latter part of the session. During qualifying, teams use minimal fuel levels in order to record the fastest lap time. The impact of fuel on lap time varies from circuit to circuit, depending on how many straights the circuit has and how long these straights are.
Tires also play a significant role in an F1 car’s performance. During each GP event, the tire supplier brings various tire types with varying compounds suitable for different racing conditions. Two of these are for wet circuit conditions: intermediate tires for light standing water and wet tires for heavy standing water. The remaining dry running tires can be categorized into three compound types: hard, medium, and soft. These tire compounds provide different grips to the circuit surface. The more grip the tire provides, the faster the car can run.
Past racing results showed that car performance dropped significantly when wet tires were used. For example, in the 2018 Italy GP, because the P1 session was wet and the qualifying session was dry, the fastest lap time in P1 was more than 10 seconds slower than the qualifying session.
Among the dry running types, the hard tire provides the least grip but is the most durable, whereas the soft tire has the most grip but is the least durable. Tires degrade over the course of a race, which reduces the tire grip and slows down the car. Track temperature and moisture affects the progression of degradation, which in turn changes the tire grip. As in the case with fuel level, tire impact on lap time changes from circuit to circuit.
Data and attempted approaches
Given this understanding of factors that can impact lap time, we can use fuel level and tire grip data to estimate the final qualifying lap time based on known practice session performance. However, as of this writing, data records to directly infer fuel level and tire grip during the race are not available. Therefore, we take an alternative approach with data we can currently obtain.
The data we used in the modeling were records of fastest lap times for each GP since 1950 and partial years of weather data for the corresponding sessions. The lap times data included the fastest lap time for each session (P1, P2, P3, and Q) of each GP with the driver, car and team, and circuit name (publicly available on F1’s website). Track wetness and temperature for each corresponding session was available in the weather data.
We explored two implicit methods with the following model inputs: the team and driver name, and the circuit name. Method one was a rule-based empirical model that attributed observed to circuits and teams. We estimated the latent parameter values (fuel level and tire grip differences specific to each team and circuit) based on their known lap time sensitivities. These sensitivities were provided by F1 and calculated through simulation runs on each circuit track. Method two was a regression model with driver and circuit indicators. The regression model learned the sensitivity of ∆t for each driver on each circuit without explicitly knowing the fuel level and tire grip exerted. We developed and compared deterministic models using XGBoost and AutoGluon, and probabilistic models using PyMC3.
We built models using race data from 2014 to 2019, and tested against race data from 2020. We excluded data from before 2014 because there were significant car development and regulation changes over the years. We removed races in which either the practice or qualifying session was wet because ∆t for those sessions were considered outliers.
Managed model training with Amazon SageMaker
We trained our regression models on Amazon SageMaker.
Amazon SageMaker is a fully managed service that provides every developer and data scientist with the ability to build, train, and deploy ML models quickly. Specifically for model training, it provides many features to assist with the process.
For our use case, we explored multiple iterations on the choices of model feature sets and hyperparameters. Recording and comparing the model metrics of interest was critical to choosing the most suitable model. The Amazon SageMaker API allowed customized metrics definition prior to launching a model training job, and easy retrieval after the training job was complete. Using the automatic model tuning feature reduced the mean squared error (MSE) metric on the test data by 45% compared to the default hyperparameter choice.
We trained an XGBoost model using the Amazon SageMaker’s built-in implementation. Its built-in implementation allowed us to run model training through a general estimator interface. This approach provided better logging, superior hyperparameter validation, and a larger set of metrics than the original implementation.
Rule-based model
In the rule-based approach, we reason that the differences of lap times ∆t primarily come from systematic variations of tire grip for each circuit and fuel level for each team between practice and qualifying sessions. After accounting for these known variations, we assume residuals are random small numbers with a mean of zero. ∆t can be modeled with the following equation:
∆tf(c) and ∆tg(c) are known sensitivities of fuel mass and tire grip, and is the residual. A hierarchy exists among the factors contained in the equation. We assume grip variations for each circuit (g(c)) are at the top level. Under each circuit, there are variations of fuel level across teams (f(t,c)).
To further simplify the model, we neglect because we assume it is small. We further assume fuel variation for each team across all circuits is the same (i.e., f(t,c) = f(t)). We can simplify the model to the following:
Because ∆tf(c) and ∆tg(c) are known, f(t) and g(c), we can estimate team fuel variations and tire grip variations from the data.
The differences in the sensitivities depend on the characteristics of circuits. From the following track maps, we can observe that the Italian GP circuit has fewer corner turns and the straight sections are longer compared to the Singapore GP circuit. Additional tire grip gives a larger advantage in the Singapore GP circuit.
ML regression model
For the ML regression method, we don’t directly model the relation between and fuel level and grip variations. Instead, we fit the following regression model with just the circuit, team, and driver indicator variables:
Ic, It, and Id represent the indicator variables for circuits, teams, and drivers.
Hierarchical Bayesian model
Another challenge with modeling the race pace was due to noisy measurements in lap times. The magnitude of random effect (ϵ) of ∆t could be non-negligible. Such randomness might come from drivers’ accidental drift from their normal practice at the turns or random variations of drivers’ efforts during practice sessions. With deterministic approaches, such random effect wasn’t appropriately captured. Ideally, we wanted a model that could quantify uncertainty about the predictions. Therefore, we explored Bayesian sampling methods.
With a hierarchical Bayesian model, we account for the hierarchical structure of the error sources. As with the rule-based model, we assume grip variations for each circuit (g(c))) are at the top level. The additional benefit of a hierarchical Bayesian model is that it incorporates individual-level variations when estimating group-level coefficients. It’s a middle ground between two extreme views of data. One extreme is to pool data for every group (circuit and driver) without considering the intrinsic variations among groups. The other extreme is to train a regression model for each circuit or driver. With 21 circuits, this amounts to 21 regression models. With a hierarchical model, we have a single model that considers the variations simultaneously at the group and individual level.
We can mathematically describe the underlying statistical model for the hierarchical Bayesian approach as the following varying intercepts model:
Here, i represents the index of each data observation, j represents the index of each driver, and k represents the index of each circuit. μjk represents the varying intercept for each driver under each circuit, and θk represents the varying intercept for each circuit. wp and wq represent the wetness level of the track during practice and qualifying sessions, and ∆T represents the track temperature difference.
Test models in the 2020 races
After predicting ∆t, we added it into the practice lap times to generate predictions of qualifying lap times. We determined the final ranking based on the predicted qualifying lap times. Finally, we compared predicted lap times and rankings with the actual results.
The following figure compares the predicted rankings and the actual rankings for all three practice sessions for the Austria, Hungary, and Great Britain GPs in 2020 (we exclude P2 for the Hungary GP because the session was wet).
For the Bayesian model, we generated predictions with an uncertainty range based on the posterior samples. This enabled us to predict the ranking of the drivers relatively with the median while accounting for unexpected outcomes in the drivers’ performances.
The following figure shows an example of predicted qualifying lap times (in seconds) with an uncertainty range for selected drivers at the Austria GP. If two drivers’ prediction profiles are very close (such as MAG and GIO), it’s not surprising that either driver might be the faster one in the upcoming qualifying session.
Metrics on model performance
To compare the models, we used mean squared error (MSE) and mean absolute error (MAE) for lap time errors. For ranking errors, we used rank discounted cumulative gain (RDCG). Because only the top 10 drivers gain points during a race, we used RDCG to apply more weight to errors in the higher rankings. For the Bayesian model output, we used median posterior value to generate the metrics.
The following table shows the resulting metrics of each modeling approach for the test P2 and P3 sessions. The best model by each metric for each session is highlighted.
MODEL | MSE | MAE | RDCG | |||
P2 | P3 | P2 | P3 | P2 | P3 | |
Practice raw | 2.822 | 1.053 | 1.544 | 0.949 | 0.92 | 0.95 |
Rule-based | 0.349 | 0.186 | 0.462 | 0.346 | 0.88 | 0.95 |
XGBoost | 0.358 | 0.141 | 0.472 | 0.297 | 0.91 | 0.95 |
AutoGluon | 0.567 | 0.351 | 0.591 | 0.459 | 0.90 | 0.96 |
Hierarchical Bayesian | 0.431 | 0.186 | 0.521 | 0.332 | 0.87 | 0.92 |
All models reduced the qualifying lap time prediction errors significantly compared to directly using the practice session results. Using practice lap times directly without considering pace correction, the MSE on the predicted qualifying lap time was up to 2.8 seconds. With machine learning methods which automatically learned pace variation patterns for teams and drivers on different circuits, we brought the MSE down to smaller than half a second. The resulting prediction was a more accurate representation of the pace in the qualifying session. In addition, the models improved the prediction of rankings by a small margin. However, there was no one single approach that outperformed all others. This observation highlighted the effect of random errors on the underlying data.
Summary
In this post, we described a new Insight developed by the Amazon ML Solutions Lab in collaboration with Formula 1 (F1).
This work is part of the six new F1 Insights powered by AWS that are being released in 2020, as F1 continues to use AWS for advanced data processing and ML modeling. Fans can expect to see this new Insight unveiled at the 2020 Turkish GP to provide predictions for the upcoming qualifying races at practice sessions.
If you’d like help accelerating the use of ML in your products and services, please contact the Amazon ML Solutions Lab .
About the Author
Guang Yang is a data scientist at the Amazon ML Solutions Lab where he works with customers across various verticals and applies creative problem solving to generate value for customers with state-of-the-art ML/AI solutions.
from AWS Machine Learning Blog https://ift.tt/3nqbL22
via A.I .Kung Fu
Black Friday Walmart deals: $199 robot vacuum, $194 AirPods Pro available now, $35 Keurig and $119 GoPro coming soon - CNET
from CNET News https://ift.tt/3plTSD4
via A.I .Kung Fu
Best Black Friday 2020 TV deals: $100 32-incher, $250 55-inch TCL, plus more soon - CNET
from CNET News https://ift.tt/3kwYCCh
via A.I .Kung Fu
Social media: How can we protect its youngest users?
from BBC News - Technology https://ift.tt/2Urn1yI
via A.I .Kung Fu
Predicting qualification ranking based on practice session performance for Formula 1 Grand Prix
If you’re a Formula 1 (F1) fan, have you ever wondered why F1 teams have very different performances between qualifying and practice sessions? Why do they have multiple practice sessions in the first place? Can practice session results actually tell something about the upcoming qualifying race? In this post, we answer these questions and more. We show you how we can predict qualifying results based on practice session performances by harnessing the power of data and machine learning (ML). These predictions are being integrated into the new “Qualifying Pace” insight for each F1 Grand Prix (GP). This work is part of the continuous collaboration between F1 and the Amazon ML Solutions Lab to generate new F1 Insights powered by AWS.
Each F1 GP consists of several stages. The event starts with three practice sessions (P1, P2, and P3), followed by a qualifying (Q) session, and then the final race. Teams approach practice and qualifying sessions differently because these sessions serve different purposes. The practice sessions are the teams’ opportunities to test out strategies and tire compounds to gather critical data in preparation for the final race. They observe the car’s performance with different strategies and tire compounds, and use this to determine their overall race strategy.
In contrast, qualifying sessions determine the starting position of each driver on race day. Teams focus solely on obtaining the fastest lap time. Because of this shift in tactics, Friday and Saturday practice session results often fail to accurately predict the qualifying order.
In this post, we introduce deterministic and probabilistic methods to model the time difference between the fastest lap time in practice sessions and the qualifying session (∆t = tq-tp). The goal is to more accurately predict the upcoming qualifying standings based on the practice sessions.
Error sources of ∆t
The delta of the fastest lap time between practice and qualifying sessions (∆t) comes primarily from variations in fuel level and tire grip.
A higher fuel level adds weight to the car and reduces the speed of the car. For practice sessions, teams vary the fuel level as they please. For the second practice session (P2), it’s common to begin with a low fuel level and run with more fuel in the latter part of the session. During qualifying, teams use minimal fuel levels in order to record the fastest lap time. The impact of fuel on lap time varies from circuit to circuit, depending on how many straights the circuit has and how long these straights are.
Tires also play a significant role in an F1 car’s performance. During each GP event, the tire supplier brings various tire types with varying compounds suitable for different racing conditions. Two of these are for wet circuit conditions: intermediate tires for light standing water and wet tires for heavy standing water. The remaining dry running tires can be categorized into three compound types: hard, medium, and soft. These tire compounds provide different grips to the circuit surface. The more grip the tire provides, the faster the car can run.
Past racing results showed that car performance dropped significantly when wet tires were used. For example, in the 2018 Italy GP, because the P1 session was wet and the qualifying session was dry, the fastest lap time in P1 was more than 10 seconds slower than the qualifying session.
Among the dry running types, the hard tire provides the least grip but is the most durable, whereas the soft tire has the most grip but is the least durable. Tires degrade over the course of a race, which reduces the tire grip and slows down the car. Track temperature and moisture affects the progression of degradation, which in turn changes the tire grip. As in the case with fuel level, tire impact on lap time changes from circuit to circuit.
Data and attempted approaches
Given this understanding of factors that can impact lap time, we can use fuel level and tire grip data to estimate the final qualifying lap time based on known practice session performance. However, as of this writing, data records to directly infer fuel level and tire grip during the race are not available. Therefore, we take an alternative approach with data we can currently obtain.
The data we used in the modeling were records of fastest lap times for each GP since 1950 and partial years of weather data for the corresponding sessions. The lap times data included the fastest lap time for each session (P1, P2, P3, and Q) of each GP with the driver, car and team, and circuit name (publicly available on F1’s website). Track wetness and temperature for each corresponding session was available in the weather data.
We explored two implicit methods with the following model inputs: the team and driver name, and the circuit name. Method one was a rule-based empirical model that attributed observed to circuits and teams. We estimated the latent parameter values (fuel level and tire grip differences specific to each team and circuit) based on their known lap time sensitivities. These sensitivities were provided by F1 and calculated through simulation runs on each circuit track. Method two was a regression model with driver and circuit indicators. The regression model learned the sensitivity of ∆t for each driver on each circuit without explicitly knowing the fuel level and tire grip exerted. We developed and compared deterministic models using XGBoost and AutoGluon, and probabilistic models using PyMC3.
We built models using race data from 2014 to 2019, and tested against race data from 2020. We excluded data from before 2014 because there were significant car development and regulation changes over the years. We removed races in which either the practice or qualifying session was wet because ∆t for those sessions were considered outliers.
Managed model training with Amazon SageMaker
We trained our regression models on Amazon SageMaker.
Amazon SageMaker is a fully managed service that provides every developer and data scientist with the ability to build, train, and deploy ML models quickly. Specifically for model training, it provides many features to assist with the process.
For our use case, we explored multiple iterations on the choices of model feature sets and hyperparameters. Recording and comparing the model metrics of interest was critical to choosing the most suitable model. The Amazon SageMaker API allowed customized metrics definition prior to launching a model training job, and easy retrieval after the training job was complete. Using the automatic model tuning feature reduced the mean squared error (MSE) metric on the test data by 45% compared to the default hyperparameter choice.
We trained an XGBoost model using the Amazon SageMaker’s built-in implementation. Its built-in implementation allowed us to run model training through a general estimator interface. This approach provided better logging, superior hyperparameter validation, and a larger set of metrics than the original implementation.
Rule-based model
In the rule-based approach, we reason that the differences of lap times ∆t primarily come from systematic variations of tire grip for each circuit and fuel level for each team between practice and qualifying sessions. After accounting for these known variations, we assume residuals are random small numbers with a mean of zero. ∆t can be modeled with the following equation:
∆tf(c) and ∆tg(c) are known sensitivities of fuel mass and tire grip, and is the residual. A hierarchy exists among the factors contained in the equation. We assume grip variations for each circuit (g(c)) are at the top level. Under each circuit, there are variations of fuel level across teams (f(t,c)).
To further simplify the model, we neglect because we assume it is small. We further assume fuel variation for each team across all circuits is the same (i.e., f(t,c) = f(t)). We can simplify the model to the following:
Because ∆tf(c) and ∆tg(c) are known, f(t) and g(c), we can estimate team fuel variations and tire grip variations from the data.
The differences in the sensitivities depend on the characteristics of circuits. From the following track maps, we can observe that the Italian GP circuit has fewer corner turns and the straight sections are longer compared to the Singapore GP circuit. Additional tire grip gives a larger advantage in the Singapore GP circuit.
ML regression model
For the ML regression method, we don’t directly model the relation between and fuel level and grip variations. Instead, we fit the following regression model with just the circuit, team, and driver indicator variables:
Ic, It, and Id represent the indicator variables for circuits, teams, and drivers.
Hierarchical Bayesian model
Another challenge with modeling the race pace was due to noisy measurements in lap times. The magnitude of random effect (ϵ) of ∆t could be non-negligible. Such randomness might come from drivers’ accidental drift from their normal practice at the turns or random variations of drivers’ efforts during practice sessions. With deterministic approaches, such random effect wasn’t appropriately captured. Ideally, we wanted a model that could quantify uncertainty about the predictions. Therefore, we explored Bayesian sampling methods.
With a hierarchical Bayesian model, we account for the hierarchical structure of the error sources. As with the rule-based model, we assume grip variations for each circuit (g(c))) are at the top level. The additional benefit of a hierarchical Bayesian model is that it incorporates individual-level variations when estimating group-level coefficients. It’s a middle ground between two extreme views of data. One extreme is to pool data for every group (circuit and driver) without considering the intrinsic variations among groups. The other extreme is to train a regression model for each circuit or driver. With 21 circuits, this amounts to 21 regression models. With a hierarchical model, we have a single model that considers the variations simultaneously at the group and individual level.
We can mathematically describe the underlying statistical model for the hierarchical Bayesian approach as the following varying intercepts model:
Here, i represents the index of each data observation, j represents the index of each driver, and k represents the index of each circuit. μjk represents the varying intercept for each driver under each circuit, and θk represents the varying intercept for each circuit. wp and wq represent the wetness level of the track during practice and qualifying sessions, and ∆T represents the track temperature difference.
Test models in the 2020 races
After predicting ∆t, we added it into the practice lap times to generate predictions of qualifying lap times. We determined the final ranking based on the predicted qualifying lap times. Finally, we compared predicted lap times and rankings with the actual results.
The following figure compares the predicted rankings and the actual rankings for all three practice sessions for the Austria, Hungary, and Great Britain GPs in 2020 (we exclude P2 for the Hungary GP because the session was wet).
For the Bayesian model, we generated predictions with an uncertainty range based on the posterior samples. This enabled us to predict the ranking of the drivers relatively with the median while accounting for unexpected outcomes in the drivers’ performances.
The following figure shows an example of predicted qualifying lap times (in seconds) with an uncertainty range for selected drivers at the Austria GP. If two drivers’ prediction profiles are very close (such as MAG and GIO), it’s not surprising that either driver might be the faster one in the upcoming qualifying session.
Metrics on model performance
To compare the models, we used mean squared error (MSE) and mean absolute error (MAE) for lap time errors. For ranking errors, we used rank discounted cumulative gain (RDCG). Because only the top 10 drivers gain points during a race, we used RDCG to apply more weight to errors in the higher rankings. For the Bayesian model output, we used median posterior value to generate the metrics.
The following table shows the resulting metrics of each modeling approach for the test P2 and P3 sessions. The best model by each metric for each session is highlighted.
MODEL | MSE | MAE | RDCG | |||
P2 | P3 | P2 | P3 | P2 | P3 | |
Practice raw | 2.822 | 1.053 | 1.544 | 0.949 | 0.92 | 0.95 |
Rule-based | 0.349 | 0.186 | 0.462 | 0.346 | 0.88 | 0.95 |
XGBoost | 0.358 | 0.141 | 0.472 | 0.297 | 0.91 | 0.95 |
AutoGluon | 0.567 | 0.351 | 0.591 | 0.459 | 0.90 | 0.96 |
Hierarchical Bayesian | 0.431 | 0.186 | 0.521 | 0.332 | 0.87 | 0.92 |
All models reduced the qualifying lap time prediction errors significantly compared to directly using the practice session results. Using practice lap times directly without considering pace correction, the MSE on the predicted qualifying lap time was up to 2.8 seconds. With machine learning methods which automatically learned pace variation patterns for teams and drivers on different circuits, we brought the MSE down to smaller than half a second. The resulting prediction was a more accurate representation of the pace in the qualifying session. In addition, the models improved the prediction of rankings by a small margin. However, there was no one single approach that outperformed all others. This observation highlighted the effect of random errors on the underlying data.
Summary
In this post, we described a new Insight developed by the Amazon ML Solutions Lab in collaboration with Formula 1 (F1).
This work is part of the six new F1 Insights powered by AWS that are being released in 2020, as F1 continues to use AWS for advanced data processing and ML modeling. Fans can expect to see this new Insight unveiled at the 2020 Turkish GP to provide predictions for the upcoming qualifying races at practice sessions.
If you’d like help accelerating the use of ML in your products and services, please contact the Amazon ML Solutions Lab .
About the Author
Guang Yang is a data scientist at the Amazon ML Solutions Lab where he works with customers across various verticals and applies creative problem solving to generate value for customers with state-of-the-art ML/AI solutions.
from AWS Machine Learning Blog https://ift.tt/3nqbL22
via A.I .Kung Fu
Hisense 70-inch 4K UHD Smart Android TV deal is still available for $400 - CNET
from CNET News https://ift.tt/3pv8Jeo
via A.I .Kung Fu