投稿時間:2021-04-15 02:32:46 RSSフィード2021-04-15 02:00 分まとめ(49件)

カテゴリー等	サイト名等	記事タイトル・トレンドワード等	リンクURL	頻出ワード・要約等/検索ボリューム	登録日
AWS	AWS News Blog	AQUA (Advanced Query Accelerator) – A Speed Boost for Your Amazon Redshift Queries	https://aws.amazon.com/blogs/aws/new-aqua-advanced-query-accelerator-for-amazon-redshift/	AQUA Advanced Query Accelerator A Speed Boost for Your Amazon Redshift QueriesAmazon Redshift already provides up to x better price performance at any scale than any other cloud data warehouse We do this by designing our own hardware and by using Machine Learning ML For example we launched the SSD based RA nodes for Amazon Redshift at the end of Amazon Redshift Update Next Generation Compute Instances …	2021-04-14 16:48:32
AWS	AWS	AQUA (Advanced Query Accelerator) for Amazon Redshift	https://www.youtube.com/watch?v=uJtCm2d5w2A	AQUA Advanced Query Accelerator for Amazon RedshiftAQUA is a new hardware accelerated cache for Amazon Redshift that enables queries to run up to x faster Learn more about AQUA at Subscribe More AWS videos ​More AWS events videos ​ AWS​ AmazonRedshift​ AQUA	2021-04-14 16:41:11
AWS	AWS	DBS Bank: Scalable Serverless Compute Grid on AWS	https://www.youtube.com/watch?v=T048vs9p1h4	DBS Bank Scalable Serverless Compute Grid on AWSIn this episode learn how DBS Bank a leading financial services group in Asia is using AWS serverless technologies to run mathematical models for trades including how they plan to scale to up to million daily executions on their serverless compute grid in AWS Check out more resources for architecting in the AWS​​​cloud ​ AWS	2021-04-14 16:06:38
AWS	AWS	Back to Basics: Secrets Management	https://www.youtube.com/watch?v=6oPHw7rT9OI	Back to Basics Secrets ManagementApplications often require sensitive information such as database credentials and API keys to operate We look at how to protect and automate the management of the secrets Additional Resources Check out more resources for architecting in the AWS cloud AWS	2021-04-14 16:05:51
Program	[全てのタグ]の新着質問一覧｜teratail（テラテイル）	tensorflowのGPU運用について	https://teratail.com/questions/333299?rss=all	tensorflowのGPU運用について前提・実現したいことAnaconda上でtensorflowをgpu運用できるようにしたいです。	2021-04-15 01:56:17
Linux	Ubuntuタグが付けられた新着投稿 - Qiita	ESPnet2をUbuntuマシンにインストールする方法	https://qiita.com/HOLY0523K/items/ee27e9fb49a35ee29c8f	ESPnetをAnaconda環境で使う場合ここにあるPytorchのどれかをインストールするようになっています。	2021-04-15 01:14:28
Git	Gitタグが付けられた新着投稿 - Qiita	巨大なリポジトリの git clone に失敗する場合の対処法	https://qiita.com/shionit/items/533a2c317fc92af73854	cdStepでcloneした巨大なリポジトリ残りのコミットログをすべて取得する残りのコミットログが一気に取得できるサイズであれば、このコマンド一発で済みますが、そもそもリポジトリ全体をgitcloneできなかったような巨大リポジトリなので、これを先に実行しても同様のエラーになると思います。	2021-04-15 01:22:14
海外TECH	Ars Technica	Verizon, AT&T, and T-Mobile kill their cross-carrier RCS messaging plans	https://arstechnica.com/?p=1756913	carrier	2021-04-14 16:28:15
海外TECH	DEV Community	Toggle dark/light mode by clapping your hands	https://dev.to/devdevcharlie/toggle-dark-light-mode-by-clapping-your-hands-li7	Toggle dark light mode by clapping your handsPost originally published on my blogA few days ago we released dark mode in beta on the Netlify app While we re still iterating on it it is currently only available via Netlify labs and then under the user s settings Even though it is not that many clicks to get there it felt like too many to me and while waiting to have it easily accessible in the nav I wondered Wouldn t it be fun to be able to toggle it by clapping your hands just like some home automation devices So I spent a part of the weekend looking into how to build it and ended up with a Chrome extension using TensorFlow js and a model trained with samples of me clapping my hands that toggles dark mode on off Here is the result FUN Or at least to me So here is how I went about it Training the modelFirst I trained the model to make sure detecting sounds of clapping hands would work To make it way easier and faster I used the Teachable machine platform and more specifically started an audio project The first section is to record what your current background sounds like Then you can create multiple classes to record samples of sounds for activities you would like to use in your project in this case clapping hands In general I record samples for a few more classes if I know that I m gonna use this model in an environment where there will be more kinds of sounds For example I also recorded samples of speech so the model would be able to recognise what speech looks like and not mistake it for clapping hands Once you re done recording your samples you can move on to training your model preview that it works as expected and export it to use it in your code Building the Chrome extensionI m not gonna go into too much detail about how to build a Chrome extension because there are a lot of different options but here s what I did for mine You need at least a manifest json file that will contain details about your extension name Dark mode clap extension description Toggle dark mode by clapping your hands version manifest version permissions storage activeTab action default icon images dark mode png images dark mode png images dark mode png images dark mode png icons images dark mode png images dark mode png images dark mode png images dark mode png content scripts js content js matches all frames true The most important parts for this project is the permissions and content scripts Content scriptContent scripts are files that run in the context of web pages so they have access to the pages the browser is currently visiting Depending on your configs in the manifest json file this would trigger on any tab or only specific tabs As I added the parameter matches this only triggers when I m on the Netlify app Then I can start triggering the code dedicated to the sound detection Setting up TensorFlow js to detect soundsWhen working with TensorFlow js I usually export the model as a file on my machine but this time I decided to use the other option and upload it to Google Cloud This way it s accessible via a URL For the code sample an example is provided on Teachable machine when you export your model but basically you need to start by creating your model const URL SPEECH MODEL TFHUB URL URL of your model uploaded on Google Cloud const recognizer await createModel async function createModel const checkpointURL URL model json const metadataURL URL metadata json model metadata const recognizer speechCommands create BROWSER FFT undefined checkpointURL metadataURL await recognizer ensureModelLoaded return recognizer And once it s done you can start the live prediction const classLabels recognizer wordLabels An array containing the classes trained In my case Background noise Clap Speech recognizer listen result gt const scores result scores will be an array of floating point numbers between and representing the probability for each class const predictionIndex scores indexOf Math max scores get the max value in the array because it represents the highest probability const prediction classLabels predictionIndex Look for this value in the array of trained classes console log prediction includeSpectrogram false probabilityThreshold invokeCallbackOnNoiseAndUnknown true overlapFactor If everything works well when this code runs it should log either Background noise Clap or Speech based on what is predicted from live audio data Now to toggle Netlify s dark mode I replaced the console log statement with some small logic The way dark mode is currently implemented is by adding a tw dark class on the body if prediction Clap if document body classList contains tw dark document body classList remove tw dark localStorage setItem nf theme light else document body classList add tw dark localStorage setItem nf theme dark I also update the value in localStorage so it is persisted Install the extensionTo be able to test that this code works you have to install the extension in your browser Before doing so you might have to bundle your extension depending on what tools you used To install it the steps to follow are Visit chrome extensionsToggle Developer mode located at the top right of the pageClick on Load unpacked and select the folder with your bundled extensionIf all goes well you should visit whatever page you want to run your extension on and it should ask for microphone permission to be able to detect live audio and start predicting That s it Overall this project wasn t even really about toggling dark mode but I ve wanted to learn about using TensorFlow js in a Chrome extension for a while so this seemed like the perfect opportunity If you want to check out the code here s the repo I can now tick that off my never ending list	2021-04-14 16:39:17
海外TECH	DEV Community	JavaScript Basic Concepts that are important in React	https://dev.to/kiranrajvjd/javascript-basic-concepts-that-are-important-in-react-3icn	JavaScript Basic Concepts that are important in ReactWhen I start learning React I was baffled by some of the codes blocks I don t understand how the code works or how the output was generated So I begin to dig into react ecosystem to understand the concepts and I found myself in no man s land I was lost and frustrated The reason I was not thorough in my JavaScript still not basics I skipped some topics believing that I fully understand the working Due to the lack of knowledge in JavaScript basics I fell into many awkward situations not knowing how to solve certain problem or what method I should use I lost too much time and I don t want anyone to have the same experience as mine So I like you to study at least the following topics in JavaScript before you jump into React ClassesModulesRest spreadDestructuringHigher order functionsArrow functionsThis list is not the ultimate list just the few topic I found most important All topic are important but try to understand at least the following which will definitely help you I have linked post that I write about the topic it may help you Some of the topics are not completing as I am trying to learn more about those topic and will be updated I would be happy if you point out my mistakes which will help me to learn better If you have any resources please feel free to share Let s learn together Happy coding	2021-04-14 16:31:31
海外TECH	DEV Community	25 Awesome Fonts From Google Fonts	https://dev.to/kiranrajvjd/25-awesome-fonts-from-google-fonts-40im	Awesome Fonts From Google FontsHere I list awesome fonts from google fonts these are my favourite fonts that I used in various projects Take a look at it and comment your favourite fonts If you have any other suggestions please comment it so that in future projects I can use those fonts AlegreyaVollkornPoppinsTitillium WebRobotoUbuntuVarelaCrimson TextBioRhymeKarlaLatoPlayfair DisplayMontserratRubikCardoNunitoOxygenMerriweatherExo AmaranthJosefin SlabArvoLoraQuicksandCairo	2021-04-14 16:17:24
海外TECH	DEV Community	Welcome Thread - v120	https://dev.to/thepracticaldev/welcome-thread-v120-5d66	Welcome Thread v Welcome to DEV Leave a comment below to introduce yourself You can talk about what brought you here what you re learning or just a fun fact about yourself Reply to someone s comment either with a question or just a hello Great to have you in the community	2021-04-14 16:06:43
Apple	AppleInsider - Frontpage News	Apple TV+, Skydance Animation announce animated short film 'Blush'	https://appleinsider.com/articles/21/04/14/apple-tv-skydance-animation-announce-animated-short-film-blush?utm_medium=rss	Apple TV Skydance Animation announce animated short film x Blush x Apple TV has announced Blush its first animated short film produced in partnership with Skydance Animation Credit AppleThe short animated film is debuting as the first piece of content in a multi year partnership between Skydance and Apple Read more	2021-04-14 16:18:49
Apple	AppleInsider - Frontpage News	How to update your medical ID in Health	https://appleinsider.com/articles/21/04/14/how-to-update-your-medical-id-in-health?utm_medium=rss	How to update your medical ID in HealthIt s been seven years since Apple introduced the Medical ID on your iPhone and you still haven t set it up Here s how you do it ーand why you should Medical ID on iPhone Read more	2021-04-14 16:16:18
海外TECH	Engadget	Unity will add native NVIDIA DLSS support to its game engine	https://www.engadget.com/unity-nvidia-dlss-support-game-engine-164535941.html	acuity	2021-04-14 16:45:35
海外TECH	Engadget	What would the internet look like without third-party cookies?	https://www.engadget.com/third-party-cookies-privacy-ad-tracking-data-floc-swan-explainer-163050682.html	What would the internet look like without third party cookies With recent news about Google committing to disabling third party cookies in Chrome it seems like the internet as we know it is about to undergo a seismic shift ーat least behind the scenes What will it look like	2021-04-14 16:30:50
海外TECH	Engadget	AT&T will launch a 'digital learning platform' with WarnerMedia content	https://www.engadget.com/att-warnermedia-digital-learning-platform-digital-divide-161523002.html	warnermedia	2021-04-14 16:15:23
海外科学	NYT > Science	How the Largest Animals That Could Ever Fly Supported Giraffe-Like Necks	https://www.nytimes.com/2021/04/14/science/pterosaurs-necks-azhdarchids.html	necksthese	2021-04-14 16:33:44
海外科学	NYT > Science	Vacuna Johnson & Johnson: Estados Unidos pausa su aplicación	https://www.nytimes.com/es/2021/04/13/espanol/coagulos-vacuna-johnson-johnson.html	extremadamente	2021-04-14 16:09:52
海外科学	NYT > Science	U.S. Calls for Pause on Johnson & Johnson Vaccine After Blood Clotting Cases	https://www.nytimes.com/2021/04/13/us/politics/johnson-johnson-vaccine-blood-clots-fda-cdc.html	U S Calls for Pause on Johnson amp Johnson Vaccine After Blood Clotting CasesFederal health officials called for a halt in the use of the company s coronavirus vaccine while they study serious illnesses that developed in six American women	2021-04-14 16:17:18
海外科学	NYT > Science	Vaccinated Mothers Are Trying to Give Babies Antibodies via Breast Milk	https://www.nytimes.com/2021/04/08/health/covid-vaccine-breast-milk.html	Vaccinated Mothers Are Trying to Give Babies Antibodies via Breast MilkMultiple studies show that there are antibodies in a vaccinated mother s milk This has led some women to try to restart breastfeeding and others to share milk with friends children	2021-04-14 16:17:31
金融	◇◇　保険デイリーニュース　◇◇（損保担当者必携！）	保険デイリーニュース(04/15)	http://www.yanaharu.com/ins/?p=4543	大同生命	2021-04-14 16:34:51
海外ニュース	Japan Times latest articles	Suga-Biden summit offers chance to hash out a plan on China	https://www.japantimes.co.jp/news/2021/04/14/national/suga-biden-summit/	china	2021-04-15 02:56:01
海外ニュース	Japan Times latest articles	Organizers celebrate 100 days until problem-plagued Tokyo Games	https://www.japantimes.co.jp/news/2021/04/14/national/japan-100-days-olympics-tokyo/	Organizers celebrate days until problem plagued Tokyo GamesCountless controversial elements regarding restrictions on spectator attendance and the priority vaccination of competing athletes among others remain undecided	2021-04-15 02:36:18
海外ニュース	Japan Times latest articles	Fukushima water release to be key issue in general election	https://www.japantimes.co.jp/news/2021/04/14/national/fukushima-water-election/	Fukushima water release to be key issue in general electionOpposition parties aim to take the water disposal question to the election for the House of Representatives which is set to take place by autumn	2021-04-15 02:29:49
海外ニュース	Japan Times latest articles	South Korea aims to fight Fukushima water release in world tribunal	https://www.japantimes.co.jp/news/2021/04/14/national/south-korea-fukushima/	tribunal	2021-04-15 02:22:45
海外ニュース	Japan Times latest articles	Tepco banned from restarting its largest nuclear plant over safety flaws	https://www.japantimes.co.jp/news/2021/04/14/national/tepco-nuclear-plant-restart-ban/	Tepco banned from restarting its largest nuclear plant over safety flawsThe company has seen restarting the seven reactor Kashiwazaki Kariwa complex once one of the world s largest nuclear plants by output as a main pillar of its	2021-04-15 02:20:12
海外ニュース	Japan Times latest articles	Top Japan virus expert acknowledges fourth wave as variants take hold	https://www.japantimes.co.jp/news/2021/04/14/national/fourth-wave-omi-coronavirus/	Top Japan virus expert acknowledges fourth wave as variants take hold Due in part to the impact of variant strains now is the time to flexibly take quasi emergency measures said Shigeru Omi who heads a government	2021-04-15 01:16:25
海外ニュース	Japan Times latest articles	Could North Korea rain missiles on Suga and Biden’s parade?	https://www.japantimes.co.jp/news/2021/04/14/national/suga-biden-summit-north-korea-missiles/	anniversary	2021-04-15 01:11:54
ニュース	BBC News - Home	Greensill: Tories reject Labour plan for MP-led lobbying probe	https://www.bbc.co.uk/news/uk-politics-56743426	starmer	2021-04-14 16:30:33
ニュース	BBC News - Home	Daunte Wright shooting: Officer Kim Potter to be charged over killing	https://www.bbc.co.uk/news/world-us-canada-56752821	degree	2021-04-14 16:36:01
ニュース	BBC News - Home	Covid-19: Care home staff in England may have to get jab, and long airport queues	https://www.bbc.co.uk/news/uk-56744272	coronavirus	2021-04-14 16:27:26
ニュース	BBC News - Home	Bernie Madoff: Disgraced financier dies in prison	https://www.bbc.co.uk/news/business-56750103	ponzi	2021-04-14 16:48:01
ニュース	BBC News - Home	Covid jab could be required for England care home staff	https://www.bbc.co.uk/news/uk-56750679	thresholds	2021-04-14 16:21:23
ニュース	BBC News - Home	Hundreds lose job in British Gas contracts row	https://www.bbc.co.uk/news/business-56746656	contracts	2021-04-14 16:52:21
ニュース	BBC News - Home	Prince Philip: Rehearsals take place for Duke of Edinburgh's funeral	https://www.bbc.co.uk/news/uk-56746947	covid	2021-04-14 16:15:20
ニュース	BBC News - Home	How many cases are there in my area?	https://www.bbc.co.uk/news/uk-51768274	explore	2021-04-14 16:52:23
Azure	Azure の更新情報	Cognitive Services - New Computer Vision API v3.2 now generally available	https://azure.microsoft.com/ja-jp/updates/cognitive-services-new-computer-vision-api-v32-now-generally-available/	Cognitive Services New Computer Vision API v now generally availableUpdated Computer Vision API now generally available to improve image tagging content moderation OCR language expansion and more	2021-04-14 16:02:07
Azure	Azure の更新情報	General availability: Log analytics workspace name uniqueness is now per resource group	https://azure.microsoft.com/ja-jp/updates/general-availability-log-analytics-workspace-name-uniqueness-is-now-per-resource-group/	General availability Log analytics workspace name uniqueness is now per resource groupYou can now use the same workspace name in deployments across all your environment without a conflict This is useful in template deployments when the same name can be used for every deployment for consistency	2021-04-14 16:02:07
Azure	Azure の更新情報	General availability: New Azure Policy built-in definitions for data encryption in Azure Monitor	https://azure.microsoft.com/ja-jp/updates/general-availability-new-azure-policy-builtin-definitions-for-data-encryption-in-azure-monitor/	General availability New Azure Policy built in definitions for data encryption in Azure MonitorWith Azure Monitor built in policy definitions for data encryption you can enforce organizational standards and assess compliance of data encryption settings in your environment	2021-04-14 16:02:04
Azure	Azure の更新情報	Public preview: Functions upgrade in Azure Monitor log analytics	https://azure.microsoft.com/ja-jp/updates/public-preview-functions-upgrade-in-azure-monitor-log-analytics/	analytics	2021-04-14 16:01:53
Azure	Azure の更新情報	Public preview: Cognitive Services- Anomaly Detector now supports multivariate anomaly detection​	https://azure.microsoft.com/ja-jp/updates/public-preview-cognitive-services-anomaly-detector-now-supports-multivariate-anomaly-detection/	Public preview Cognitive Services Anomaly Detector now supports multivariate anomaly detection​Enable organizations to identify anomalies across multiple variables with multivariate Anomaly Detector	2021-04-14 16:01:46
Azure	Azure の更新情報	Reduce dataflow execution time with cluster reuse public preview in Azure Data Factory	https://azure.microsoft.com/ja-jp/updates/reduce-dataflow-execution-time-with-cluster-reuse-public-preview-in-azure-data-factory/	Reduce dataflow execution time with cluster reuse public preview in Azure Data FactoryAzure Data Factory ADF has released a “quick re use option as public preview to the Azure Integration Runtime TTL to reduce data flow execution to from mins to under seconds	2021-04-14 16:01:39
Azure	Azure の更新情報	Public preview of Microsoft Build of OpenJDK	https://azure.microsoft.com/ja-jp/updates/public-preview-of-microsoft-build-of-openjdk/	build	2021-04-14 16:01:32
Azure	Azure の更新情報	Public preview of SAP NetWeaver, North Europe, and new updates in cluster monitoring	https://azure.microsoft.com/ja-jp/updates/sap-on-azure-public-preview-of-sap-netweaver-in-north-europe/	Public preview of SAP NetWeaver North Europe and new updates in cluster monitoringAzure Monitor for SAP Solutions AMS supports SAP NetWeaver monitoring in public preview new metrics for high availability pacemaker clusters and availability in North Europe region	2021-04-14 16:01:25
Azure	Azure の更新情報	Azure IoT Edge nesting capabilities are now generally available	https://azure.microsoft.com/ja-jp/updates/azure-iot-edge-nesting-capabilities-are-now-generally-available/	Azure IoT Edge nesting capabilities are now generally availableOur commitment is to simplify IoT for mainstream adoption As such we are announcing general availability of nesting capabilities for Azure IoT Edge for industrials to connect their equipment to the cloud through multiple network layers as recommended by the ANSI ISA standard	2021-04-14 16:01:18
Azure	Azure の更新情報	Azure SQL—general availability updates for April 14, 2021	https://azure.microsoft.com/ja-jp/updates/azure-sql-general-availability-updates-for-april-14-2021/	april	2021-04-14 16:01:13
Azure	Azure の更新情報	Azure Database for PostgreSQL – Hyperscale (Citus) same region read replicas in public preview	https://azure.microsoft.com/ja-jp/updates/azure-database-for-postgresql-hyperscale-citus-same-region-read-replicas-in-public-preview/	group	2021-04-14 16:01:07
GCP	Cloud Blog	Churn prediction for game developers using Google Analytics 4 (GA4) and BigQuery ML	https://cloud.google.com/blog/topics/developers-practitioners/churn-prediction-game-developers-using-google-analytics-4-ga4-and-bigquery-ml/	Churn prediction for game developers using Google Analytics GA and BigQuery MLUser retention can be a major challenge for mobile game developers According to the Mobile Gaming Industry Analysis in most mobile games only see a retention rate for users after the first day To retain a larger percentage of users after their first use of an app developers can take steps to motivate and incentivize certain users to return But to do so developers need to identify the propensity of any specific user returning after the first hours  In this blog post we will discuss how you can use BigQuery ML to run propensity models on Google Analytics data from your gaming app to determine the likelihood of specific users returning to your app You can also use the same end to end solution approach in other types of apps using Google Analytics for Firebase as well as apps and websites using Google Analytics To try out the steps in this blogpost or to implement the solution for your own data you can use this Jupyter Notebook  Using this blog post and the accompanying Jupyter Notebook you ll learn how to Explore the BigQuery export dataset for Google Analytics Prepare the training data using demographic and behavioural attributesTrain propensity models using BigQuery MLEvaluate BigQuery ML modelsMake predictions using the BigQuery ML modelsImplement model insights in practical implementationsGoogle Analytics GA properties unify app and website measurement on a single platform and are now default in Google Analytics Any business that wants to measure their website app or both can use GA for a more complete view of how customers engage with their business With the launch of Google Analytics BigQuery export of Google Analytics data is now available to all users If you are already using a Google Analytics property you can follow this guide to set up exporting your GA data to BigQuery Once you have set up the BigQuery export you can explore the data in BigQuery Google Analytics uses an event based measurement model Each row in the data is an event with additional parameters and properties The Schema for BigQuery Export can help you to understand the structure of the data   In this blogpost we use the public sample export data from an actual mobile game app called Flood It Android iOS to build a churn prediction model But you can use data from your own app or website  Here s what the data looks like Each row in the dataset is a unique event which can contain nested fields for event parameters This dataset contains M events from over k users Our goal is to use BigQuery ML on the sample app dataset to predict propensity to user churn or not churn based on users demographics and activities within the first hours of app installation In the following sections we ll cover how to Pre process the raw event data from GAIdentify users amp the label featureProcess demographic featuresProcess behavioral featuresTrain classification model using BigQuery MLEvaluate the model using BigQueryMLMake predictions using BigQuery MLUtilize predictions for activationPre process the raw event dataYou cannot simply use raw event data to train a machine learning model as it would not be in the right shape and format to use as training data So in this section we ll go through how to pre process the raw data into an appropriate format to use as training data for classification models This is what the training data should look like for our use case at the end of this section Notice that in this training data each row represents a unique user with a distinct user ID user pseudo id  Identify users amp the label featureWe first filtered the dataset to remove users who were unlikely to return the app anyway We defined these bounced users as ones who spent less than mins with the app Then we labeled all remaining users churned No event data for the user after hours of first engaging with the app returned The user has at least one event record after hours of first engaging with the app For your use case you can have a different definition of bounce and churning Also you can even try to predict something else other than churning e g whether a user is likely to spend money on in game currency likelihood of completing n number of game levelslikelihood of spending n amount of time in game etc In such cases label each record accordingly so that whatever you are trying to predict can be identified from the label column From our dataset we found that users bounced However from the remaining users   churned after hours To create these bounced and churned columns we used the following snippet of SQL code  You can view the Jupyter Notebook for the full query used for materializing the bounced and churned labels  Process demographic featuresNext we added features both for demographic data and for behavioral data spanning across multiple columns Having a combination of both demographic data and behavioral data helps to create a more predictive model  We used the following fields for each user as demographic features geo countrydevice operating systemdevice languageA user might have multiple unique values in these fields for example if a user uses the app from two different devices To simplify we used the values from the very first user engagement event Process behavioral featuresThere is additional demographic information present in the GA export dataset e g app info device event params geo etc You may also send demographic information to Google Analytics through each hit via user properties Furthermore if you have first party data on your own system you can join that with the GA export data based on user ids  To extract user behavior from the data we looked into the user s activities within the first hours of first user engagement In addition to the events automatically collected by Google Analytics there are also the recommended events for games that can be explored to analyze user behavior For our use case to predict user churn we counted the number of times the follow events were collected for a user within hours of first user engagement  user engagementlevel start quickplaylevel end quickplaylevel complete quickplaylevel reset quickplaypost scorespend virtual currencyad rewardchallenge a friendcompleted levelsuse extra stepsThe following query shows how these features were calculated View the notebook for the query used to aggregate and extract the behavioral data You can use different sets of events for your use case To view the complete list of events use the following query After this we combined the features to ensure our training dataset reflects the intended structure We had the following columns in our table User ID user pseudo idLabel churnedDemographic featurescountrydevice osdevice languageBehavioral featurescnt user engagementcnt level start quickplaycnt level end quickplaycnt level complete quickplaycnt level reset quickplaycnt post scorecnt spend virtual currencycnt ad rewardcnt challenge a friendcnt completed levelscnt use extra stepsuser first engagementAt this point the dataset was ready to train the classification machine learning model in BigQuery ML Once trained the model will output a propensity score between churn churned or return churned indicating the probability of a user churning based on the training data Train classification model When using the CREATE MODEL statement BigQuery ML automatically splits the data between training and test Thus the model can be evaluated immediately after training see the documentation for more information For the ML model we can choose among the following classification algorithms where each type has its own pros and cons Often logistic regression is used as a starting point because it is the fastest to train The query below shows how we trained the logistic regression classification models in BigQuery ML We extracted month julianday and dayofweek from datetimes timestamps as one simple example of additional feature preprocessing before training Using TRANSFORM in your CREATE MODEL query allows the model to remember the extracted values Thus when making predictions using the model later on these values won t have to be extracted again View the notebook for the example queries to train other types of models XGBoost deep neural network AutoML Tables Evaluate modelOnce the model finished training we ran ML EVALUATE to generate precision recall accuracy and f score for the model The optional THRESHOLD parameter can be used to modify the default classification threshold of For more information on these metrics you can read through the definitions on precision and recall accuracy f score log loss and roc auc Comparing the resulting evaluation metrics can help to decide among multiple models Furthermore we used a confusion matrix to inspect how well the model predicted the labels compared to the actual labels The confusion matrix is created using the default threshold of which you may want to adjust to optimize for recall precision or a balance more information here This table can be interpreted in the following way Make predictions using BigQuery MLOnce the ideal model was available we ran ML PREDICT to make predictions For propensity modeling the most important output is the probability of a behavior occurring The following query returns the probability that the user will return after hrs The higher the probability and closer it is to the more likely the user is predicted to return and the closer it is to the more likely the user is predicted to churn Utilize predictions for activationOnce the model predictions are available for your users you can activate this insight in different ways In our analysis we used user pseudo id as the user identifier However ideally your app should send back the user id from your app to Google Analytics In addition to using first party data for model predictions this will also let you join back the predictions from the model into your own data You can import the model predictions back into Google Analytics as a user attribute This can be done using the Data Import feature for Google Analytics Based on the prediction values you can Create and edit audiences and also do Audience targeting For example an audience can be users with prediction probability between and to represent users who are predicted to be on the fence between churning and returning For Firebase Apps you can use the Import segments feature You can tailor user experience by targeting your identified users through Firebase services such as Remote Config Cloud Messaging and In App Messaging This will involve importing the segment data from BigQuery into Firebase After that you can send notifications to the users configure the app for them or follow the user journeys across devices Run targeted marketing campaigns via CRMs like Salesforce e g send out reminder emails You can find all of the code used in this blogpost in the Github repository What s next  Continuous model evaluation and re trainingAs you collect more data from your users you may want to regularly evaluate your model on fresh data and re train the model if you notice that the model quality is decaying Continuous evaluationーthe process of ensuring a production machine learning model is still performing well on new dataーis an essential part in any ML workflow Performing continuous evaluation can help you catch model drift a phenomenon that occurs when the data used to train your model no longer reflects the current environment  To learn more about how to do continuous model evaluation and re train models you can read the blogpost Continuous model evaluation with BigQuery ML Stored Procedures and Cloud SchedulerMore resourcesIf you d like to learn more about any of the topics covered in this post check out these resources BigQuery export of Google Analytics dataBigQuery ML quickstartEvents automatically collected by Google Analytics Qwiklabs Create ML models with BigQuery MLOr learn more about how you can use BigQuery ML to easily build other machine learning solutions How to build demand forecasting models with BigQuery MLHow to build a recommendation system on e commerce data using BigQuery MLLet us know what you thought of this post and if you have topics you d like to see covered in the future You can find us on Twitter at polonglin and mkazi Thanks to reviewers Abhishek Kashyap Breen Baker  David Sabater Dinter Related ArticleHow to build demand forecasting models with BigQuery MLWith BigQuery ML you can train and deploy machine learning models using SQL With the fully managed scalable infrastructure of BigQuery Read Article	2021-04-14 16:30:00