How to Implement Machine Learning Projects with Your Data and AutoML [HOL2948].pdf

《How to Implement Machine Learning Projects with Your Data and AutoML [HOL2948].pdf》由會員分享，可在線閱讀，更多相關《How to Implement Machine Learning Projects with Your Data and AutoML [HOL2948].pdf（87頁珍藏版）》請在三個皮匠報告上搜索。

1、How to Implement Machine Learning Projects with Your Data and AutoMLMarcos Arancibia,Mark HornickSeptember 10,2024Session code HOL2948Oracle CloudWorld Copyright 2024,Oracle and/or its affiliates1AgendaOracle CloudWorld Copyright 2024,Oracle and/or its affiliates2 Welcome and Introduction Overview o

2、f todays Lab Introduction to the workshop environment Accessing and using Oracle Machine Learning Wrap up and next stepsWelcome and IntroductionHow to implement machine learning projects with your data and AutoMLOracle CloudWorld Copyright 2024,Oracle and/or its affiliates3MarkHornickSr Director of

3、Product MgtOracle Machine LearningMarcos ArancibiaSr Principal Product ManagerAutonomous DatabaseOracle CloudWorld Copyright 2024,Oracle and/or its affiliates4The teamAgendaOracle CloudWorld Copyright 2024,Oracle and/or its affiliates5Introduction to ML concepts and the workshopGet Started with Live

4、Labs Provision a LiveLabs environment with Autonomous Data Warehouse and an OML-enabled userTask 1 Access Oracle Machine Learning UITask 2 Use AutoML UI to build a model that predicts Customer churnTask 3 Evaluate and deploy the ModelTask 4 Load a Notebook that uses the model for scoringTask 5 Let A

5、utoML build a Notebook Task 6 Deploy the model to a REST endpointIntroduction to todays workshopA fictional company named MovieStream(media streaming)is losing customers to competitors.You are tasked to identify at-risk customers to send to the Dev team.Developers will then be able to present specia

6、l offers to increase loyaltyOracle CloudWorld Copyright 2024,Oracle and/or its affiliates6https:/youtu.be/C0DWnTpSL-kIntroducing Oracle MovieStreamOracle CloudWorld Copyright 2024,Oracle and/or its affiliates7There is a lot of intelligence behind the applicationOracle MovieStream:a personalized user

7、 experienceMachine Learning+Spatial AnalyticsPredict churn and localize an offerAnalytic SQLRankings over recent transactionsAnalytic SQL+SpatialRankings refined based on locationJSON AnalyticsQuery complex JSON data typesGraph AnalyticsRecommendations using sophisticated graph algorithmsOracle Clou

8、dWorld Copyright 2024,Oracle and/or its affiliates8Oracle CloudWorld Copyright 2024,Oracle and/or its affiliates9Developing the personalized experienceFront-end developer viewpointMust deliver a compelling user experienceNeed a productive,easy to use API plus fast,reliable performanceDont know and d

9、ont care how page contents are derived!Oracle CloudWorld Copyright 2024,Oracle and/or its affiliates10Developing the personalized experienceFront-end developer viewpointMust deliver a compelling user experienceNeed a productive,easy to use API plus fast,reliable performanceDont know and dont care ho

10、w page contents are derived!Simplify data with Autonomous DatabaseDo more with dataPut your data to work using built-in self-service tools and your favorite applications.Focus on innovating and gaining insights from all your data.Sleep well at night with the confidence that your data is always safe

11、and available.02IntegrateCollect and wrangle data setsOperationalizeApply insights to improve processes and mitigate risks01DiscoverFind and understand data sets available across the enterprise03AnalyzeUse your favorite applications to develop actionable insights04Oracle CloudWorld Copyright 2024,Or

12、acle and/or its affiliates11Introduction to the workshop environmentHow to implement machine learning projects with your data and AutoMLOracle CloudWorld Copyright 2024,Oracle and/or its affiliates12Showcasing how Oracles solutions solve your business problems available 24/7 for free!13Oracle CloudW

13、orld Copyright 2024,Oracle and/or its affiliatesOracle LiveLabs1000+events run using LiveLabsworkshops700+free workshops,available or in development10 millionpeople have already visited LiveLabsLaptopOracle AccountRequirements for running a Hands-on-LabOracle CloudWorld Copyright 2024,Oracle and/or

14、its Hands-on-Lab15Oracle CloudWorld Copyright 2024,Oracle and/or its a web browser2.Navigate to the provided URL3.Use your Oracle SSO account to login Please note:This is NOT your Oracle Cloud Infrastructure accountOracle CloudWorld Copyright 2024,Oracle and/or its affiliates16Lets get the Start but

15、ton to access the HOL2.Click Run on LiveLabs Sandbox3.Toggle the Start Workshop Now radio button4.Check the consent box5.Click Submit ReservationOracle CloudWorld Copyright 2024,Oracle and/or its affiliates17Lets get Launch Workshop once availablePlease note:The environments are available in a coupl

16、e of minutes.Refresh the pageOracle CloudWorld Copyright 2024,Oracle and/or its affiliates18Lets get startedIntroduction to Machine Learning conceptsHow to implement machine learning projects with your data and AutoMLOracle CloudWorld Copyright 2024,Oracle and/or its affiliates19Requirements for an

17、effective Machine Learning project in 2024 1.Need agility to build Machine Learning models,in many languages Need to support Data Scientists and their popular APIs for ML2.Require a flexible environment with self-service tools Avoid duplication and transfer of data between different tools and platfo

18、rms3.Must be simple to deploy ML models to production You need the ability to deploy models easily for consumption by any Applications,using several APIsOracle CloudWorld Copyright 2024,Oracle and/or its affiliates20Machine Learning use cases are part of every industryCustomers Segmentation Loyalty

19、Lifetime value Retention/churn Acquisition Lead generation and prioritization Factors of best customers Invoice anomaly detection Fraud detectionProducts Next best offer Cross-sell Upsell Product bundling Demand forecasting Inventory forecasting Revenue predictionEquipment Predictive maintenance Roo

20、t cause analysis Demand forecasting Usage anomaly detectionEmployees Attributes of best employees Retention Hiring Predicting hiring needsOracle CloudWorld Copyright 2024,Oracle and/or its affiliates21Use cases and machine learning techniquesAddress important business problems that impact customers,

21、products,operations,and employeesClassificationRegressionClusteringCustomer lifetime value predictionLoan prepayment predictionProbability of defaultPredictive maintenanceCustomer loyalty/churnCustomer acquisitionEmployee retentionAssociation RulesTime SeriesProduct cross sell and upsellProduct bund

22、lingDemand forecastingSales and revenue forecastingATM withdrawal forecastingCustomer segmentationDocument classificationBiological species classificationLocation-based house value analysisExploratory data analysisNext best offerAnomaly DetectionFraud detectionNetwork intrusion detectionUnusual case

23、 identificationOracle CloudWorld Copyright 2024,Oracle and/or its affiliates22Instead of single-use proprietary databases run converged,open DatabaseOracles Converged DatabaseMachine LearningSpatialGraphJSONBlockchainAll modern data types,analytics,and the latest development paradigms built into one

24、 product at no additional costRESTIoTIn-memoryVectorOracle CloudWorld Copyright 2024,Oracle and/or its affiliates23Oracle Machine Learning family of componentsOML ComponentAutonomous DatabaseServerless|DRAutonomous Database Dedicated|CCOracle Database on premises,BDBS,Exadata CS|CI|CCOML4SQL APIBuil

25、d ML models and score data with no data movement using SQL and PL/SQL OML4Py APILeverage the database as a high-performance compute engine from Python with in-database ML OML4R APILeverage the database as a high-performance compute engine from R with in-database ML OML NotebooksSQL,PL/SQL,Python,R,c

26、onda,and markdown interpreters OML AutoML UINo-code automated modeling interface OML ServicesRESTful model management,deployment,monitoring Oracle Data MinerSQL Developer extension with a drag-n-drop interface for creating ML methodologies Oracle CloudWorld Copyright 2024,Oracle and/or its affiliate

27、s24Oracle Machine LearningMachine learning algorithms in the database kernel software Eliminate data movement Simplify solution architectures Speed solution development and deployment Augment applications and dashboards quickly and easilyMove the algorithms not the dataOracle Machine Learning databa

28、se features included with Oracle Database and Oracle Autonomous DatabaseOver 30 scalable in-database ML algorithms using SQL,R and Python APIs and automated no-code user interfacesML functionality in Oracles converged database with spatial,graph,blockchain,JSONEasily use with OAS/OAC,Oracle APEX,OCI

29、 DS,ODI,and othersOracle CloudWorld Copyright 2024,Oracle and/or its affiliates25Address a wide range of business problemsOracle Machine Learning In-Database ML AlgorithmsClassificationRegressionAnomaly DetectionClusteringTime SeriesAttribute ImportanceFeature ExtractionRow ImportanceAssociation Rul

30、esRankingSurvival AnalysisDecision TreeExplicit Semantic AnalysisLogistic Regression(GLM)Nave BayesNeural NetworkRandom ForestSupport Vector Machine(SVM)XGBoostGeneralized Linear Model(GLM)Neural NetworkSupport Vector Machine(SVM)Stepwise Linear regressionXGBoostHierarchical K-MeansHierarchical O-Cl

31、usterExpectation MaximizationExponential SmoothingMultiple Time Series(23c)Includes popular models e.g.Holt-Winters with trends,seasonality,irregular time seriesPrincipal Comp Analysis(PCA)Non-negative Matrix FactorizationSingular Value Decomposition(SVD)Explicit Semantic Analysis(ESA)Minimum Descri

32、ption LengthRandom ForestUnsupervised Pairwise KL Divergence CUR decomposition for row&AIA priori CUR DecompositionXGBoostXGBoostOne-Class SVMMSET-SPRTExpectation Maximization(23c)Includes support for partitioned models,integrated text mining,automatic data preparation OML Algorithm Cheat Sheet Algo

33、rithm Documentation OML Performance on ADBOracle CloudWorld Copyright 2024,Oracle and/or its affiliates26The problem with traditional Machine Learning ModelingOracle CloudWorld Copyright 2024,Oracle and/or its affiliates27ModelingGLMStep 1Step 2Step 3U()DTStep 1Step 2Step 3NNStep 1Step 2Step 3SVMSte

34、p 1Step 2Step 3SVMStep 1Step 2Step 3DataML ProcessDecision TreeGeneralized Linear ModelNave BayesNeural NetworkRandom ForestSupport Vector MachineCOMPLEXITY_FACTORCONV_TOLERANCEEPSILONKERNEL_FUNCTIONOUTLIER_RATESTD_DEVNUM_ITERATIONSNUM_PIVOTSBATCH_ROWSREGULARIZERSOLVER27Copyright 2023,Oracle and/or

35、its affiliates NNStep 1Step 2Step 3An example of building a Churn Prediction model using OML AutoML UIEliminate repetitive tasks of model building/evaluation to increase user productivityEnable non-expert users to leverage machine learningApply ML to the ML process to reduce algorithm and hyperparam

36、eters search space and reduce compute time and costOMLModelAuto Algorithm SelectionIdentify in-database algorithms likely to achieve highermodel quality Find best algorithm faster than exhaustive searchAdaptive SamplingIdentify right sample size for training dataAdjust sample for unbalanced dataAuto

37、 Feature SelectionDe-noise dataReduce features by identifying most predictiveImprove accuracy and performance Auto Model TuningImproves model accuracyAutomated tuning of hyperparametersAvoid manual or exhaustive search techniquesOML AutoML UI Experiment PipelineFeature Prediction ImpactRank features

38、 most influential for scoringAlgorithm-agnostic techniqueFor each final model per algorithmPlusDataOracle CloudWorld Copyright 2024,Oracle and/or its affiliates28Oracle CloudWorld Copyright 2024,Oracle and/or its affiliates29Click your way to an ML modelEnhance data scientist productivity and help n

39、on-experts produce ML modelsSelect the prepared data table and the column you want to predictStart automated build and compare of multiple models with model quality metricsGenerate editable notebooks for desired models with AutoML-selected hyperparameter valuesRename models to easily recognize model

40、s in model repositoryDeploy models immediately using SQL or deploy to OML Services as REST endpointsTask:Navigate to Oracle Machine Learning UIHow to implement machine learning projects with your data and AutoMLOracle CloudWorld Copyright 2024,Oracle and/or its View Login Info2.Click Copy Value to c

41、opy the Database User Password3.Click OML UI to go to the Oracle Machine Learning environmentOracle CloudWorld Copyright 2024,Oracle and/or its affiliates31Lets get started123Navigate to OML UIOracle CloudWorld Copyright 2024,Oracle and/or its affiliates32At the OML UI Sign In page,use OMLUSER,enter

42、 the password AAbbcc123456,then click Sign InOnce in the OML UI,click on AutoMLOracle CloudWorld Copyright 2024,Oracle and/or its affiliates33You can access AutoML UI by:Clicking on the three-lines menu and then select AutoML Experiments,orClicking on the AutoML button in the Quick Actions sectionTa

43、sk:Create a new AutoML ExperimentHow to implement machine learning projects with your data and AutoMLOracle CloudWorld Copyright 2024,Oracle and/or its affiliates34In the AutoML Experiments page,click“Create”Oracle CloudWorld Copyright 2024,Oracle and/or its affiliates35Give the experiment a Name an

44、d an optional CommentAfter entering a name and optional comment,Click on the magnifying glass icon to search for the input data to useOracle CloudWorld Copyright 2024,Oracle and/or its affiliates36Search for CHURN and select the table MOVIESTREAM_CHURN by clicking on it and then OKClick OK to confir

45、m your selectionOracle CloudWorld Copyright 2024,Oracle and/or its affiliates37The Create Experiment screen should look like the followingOracle CloudWorld Copyright 2024,Oracle and/or its affiliates38Lets define the Experiments target and Unique IDClick the drop-down in the Predictfield.Scroll to a

46、nd select the target attribute named IS_CHURNER.For the Case ID pull-down menu selection,choosing the column CUST_ID as the unique customer identifierOracle CloudWorld Copyright 2024,Oracle and/or its affiliates39Under Additional Settings,reduce the Maximum Top Models to 3.We can optionally set the

47、Database Level to HighNOTE:Under algorithms we are not changing the defaults,which is to try to use any algorithm available in the search.Specific algorithms can be excluded from the search if necessaryOracle CloudWorld Copyright 2024,Oracle and/or its affiliates40Lets remove customer names from the

48、 model,as it does not make business sense in estimating customer likelihood to churn1.Scroll down to the Features list,type in NAME in the Search box2.Uncheck the 2 boxes that contain FIRST_NAME and LAST_NAME so they will not be part of the modelOracle CloudWorld Copyright 2024,Oracle and/or its aff

49、iliates41Task:Run the Experiment and explore the resultsHow to implement machine learning projects with your data and AutoMLOracle CloudWorld Copyright 2024,Oracle and/or its affiliates42At the Top Right of the screen,click Start and then Faster ResultsThe experiment will start runningOracle CloudWo

50、rld Copyright 2024,Oracle and/or its affiliates43The Leader Board will show the progress and best models selected.We expect it should be completed in a couple of minutes.Oracle CloudWorld Copyright 2024,Oracle and/or its affiliates44If you scroll down in to the Features list,you will see a new Impor

51、tance column,that shows general importance of the featureThis feature importance indicates the overall level of sensitivity of prediction to a particular feature,but is not necessarily related to any of the models createdOracle CloudWorld Copyright 2024,Oracle and/or its affiliates45The prediction i

52、mpacts show the most important attributes(identified by the model)that influence the targetLets now check the Model details.Click on the name of the Top model itself(its the one that starts with RF_)The most important features for this particular Random Forest model include geo-demographic attribute

53、s like AGE GENDER EDUCATIONAs well as behavioral attributes like GENRE_THRILLER(number of Thriller movies watched last 6 mos)AVG_NTRANS_M3_5(average number of transactions in the previous quarter)and others with some lesser impact,but still part of the model.Oracle CloudWorld Copyright 2024,Oracle a

54、nd/or its affiliates46Lets check the Confusion Matrix nowThe confusion matrix is computed by OML AutoML UI on a random subset of the original data.It shows how many customers the model thought were going to accept the offer(Predicted=1)or not(Predicted=0)vs.the actual results from the test set rando

55、mly selected from the experiment data.In the Test Set this particular Random Forest model was able to correctly identify 93.03%of the future customer behaviors(the green diagonals,64.81%+28.22%).The model incorrectly thought that 5.92%of the test dataset were going to churn(Predicted:1),but they act

56、ually did not(Actual:0),known as False Positives.This is actually not that bad for marketing purposes,since we would be sending offers to these customers when they might be staying with us anyway.The model also thought that 1.05%of the customers would not churn(Predict:0)but they actually did(Actual

57、:1),known as False Negatives.This type of error is more dangerous in churn management,since your model is not capable of identifying these customers as churners when they indeed would have been,and should be monitored over time.Oracle CloudWorld Copyright 2024,Oracle and/or its affiliates47Task:Prep

58、are the model for scoring and deploymentHow to implement machine learning projects with your data and AutoMLOracle CloudWorld Copyright 2024,Oracle and/or its affiliates48We will click anywhere in the same row of the top model(but not on directly its name),to select it.We will Rename the model to so

59、mething that makes business senseNOTE:clicking anywhere in the same row of the model will highlight that row in light blue,indicating that it is selectedLets rename the model CHURN_PREDOracle CloudWorld Copyright 2024,Oracle and/or its affiliates49The Leader Board should look like this nowOracle Clo

60、udWorld Copyright 2024,Oracle and/or its affiliates50Task:Score customers with Python and SQL using the modelHow to implement machine learning projects with your data and AutoMLOracle CloudWorld Copyright 2024,Oracle and/or its affiliates51Lets download the notebook indicated in the Instructions at

61、the section Score customers with Python and SQL using the model.Click the button indicated and download the file locallyOracle CloudWorld Copyright 2024,Oracle and/or its affiliates52Lets open the notebook indicated in the Instructions by clicking on the three-lines menu and selecting Notebooks.Orac

62、le CloudWorld Copyright 2024,Oracle and/or its affiliates5312Click on import to load the file Scoring Customers with Churn Model.dsnbOracle CloudWorld Copyright 2024,Oracle and/or its affiliates54The Notebook is going to be in the listing now.Click on it to open.Oracle CloudWorld Copyright 2024,Orac

63、le and/or its affiliates55The Notebook will be opened it in editing mode.Oracle CloudWorld Copyright 2024,Oracle and/or its affiliates56Lets run the Notebook and check the results.Click on the play button at the top of the screen,and confirm your choiceOracle CloudWorld Copyright 2024,Oracle and/or

64、its affiliates57After a successful run,hovering the mouse over any paragraph will show how long ago the Notebook ran lastAfter the computation is done,materialize the temporary table to the DatabaseFirst section:Using Python with OML4Py to compute churnOracle Machine Learning for Python can map Tabl

65、es and OML models from the Database without having to bring them into local memory,using proxies.It can then score them via proxies and command the database to write the physical output to the Database,so the data never has to leave the Database itself,for increased security and performanceOur objec

66、tive were these two columns:The probability of a customer to churn,and whether they will churn(prob50%)Oracle CloudWorld Copyright 2024,Oracle and/or its affiliates58Scroll down to the SQL sectionYou can use the same CHURN_PRED model to score(create inference)via SQL dynamically a table with the sam

67、e columns required for INPUT by the model.Just as example we are using MOVIESTREAM_CHURNOracle has a unique feature in SQL for in-Database Dynamic Machine Learning Inference.The functions PREDICTION()and PREDICTION_PROBABILITY()can score any table that contains the columns necessary for computing th

68、e Probability.Oracle Machine Learning is also resilient to missing data,and it can return a Probability computation even if there are missing data in some columns Oracle CloudWorld Copyright 2024,Oracle and/or its affiliates60Last SQL section shows a sample of the materialized tableOracle CloudWorld

69、 Copyright 2024,Oracle and/or its affiliates61Task:Additional model quality metricsHow to implement machine learning projects with your data and AutoMLOracle CloudWorld Copyright 2024,Oracle and/or its affiliates62Lets go back to the OML AutoML UI to see a few more featuresClick on the three-lines m

70、enu and then on AutoML ExperimentsOracle CloudWorld Copyright 2024,Oracle and/or its affiliates63Select the experiment we createdOracle CloudWorld Copyright 2024,Oracle and/or its affiliates64We will click anywhere in the same row of the top model(but not on directly its name),to select it.Then clic

71、k on Metrics,and select a few new metrics.NOTE:clicking anywhere in the same row of the model will highlight that row in light blue,indicating that it is selectedOracle CloudWorld Copyright 2024,Oracle and/or its affiliates65You will see the additional metrics next to the model namesOracle CloudWorl

72、d Copyright 2024,Oracle and/or its affiliates66Task:Auto-generated OML4Py NotebooksHow to implement machine learning projects with your data and AutoMLOracle CloudWorld Copyright 2024,Oracle and/or its affiliates67In the leaderboard,select the row for CHURN_PRED and click Create NotebookAccept the d

73、efault name and click OKAfter it is created,a pop-up note will appear at the bottom of the screen.Click Open NotebookOracle CloudWorld Copyright 2024,Oracle and/or its affiliates68You might be asked to Allow Run the markdown code,which is safe in this caseLets start by running the entire notebook.Th

74、e first section shows details of when the model was created,the username,input data,predict,algorithm and other details.Oracle CloudWorld Copyright 2024,Oracle and/or its affiliates69Scroll down-data preparation sectionOracle CloudWorld Copyright 2024,Oracle and/or its affiliates70This section is pr

75、eparing a query that will include only the columns that were found to have significance to this particular model,and then it will create a proxy object to the data.Scroll down-build an RF model with the best hyperparametersOracle CloudWorld Copyright 2024,Oracle and/or its affiliates71This section i

76、s building the input dataset and then creating the Random Forest model from scratch,showing the exact hyperparameters used to build the original model created via AutoML UI.It also allows experts to make additional adjustments.The model is built using the fit()function.Lets review the model detailsO

77、racle CloudWorld Copyright 2024,Oracle and/or its affiliates72The next section illustrates the output of the model just created,by showing the model details.Scrolling down shows many more of the details.Scoring data in Python with the new modelThe last sections show how to score(generate inference)o

78、f customers,and verify the model quality as well using the open-source sklearn package from Python.Task:Deploy the model as an OML Services REST endpointHow to implement machine learning projects with your data and AutoMLOracle CloudWorld Copyright 2024,Oracle and/or its affiliates74Lets go back to

79、the OML AutoML UI to see a few more featuresClick on the three-lines menu and then on AutoML ExperimentsOracle CloudWorld Copyright 2024,Oracle and/or its affiliates75Select the line in the Leader Board with the CHURN_PRED model and select Deploy from the menu.In the dialog that appears,type the inf

80、ormation as indicated and click OK.Lets deploy the modelOracle CloudWorld Copyright 2024,Oracle and/or its affiliates76NOTE:clicking anywhere in the same row of the model will highlight that row in light blue,indicating that it is selectedClick the three-line menu at the top of the Oracle Machine Le

81、arning,and then click Models in the menu that opens.Lets go to the Models section of OML.We are interested on the Deployment we just did,so lets go there by clicking on DeploymentsThe models we have built so far are available in the Database,and listed hereThe list of deployments is shown,and should

82、 contain only oneUnder URI,the name of the deployment is the one we gave it when deploying itThe metadata contains the details about the model,what mining function it uses,the expected inputs that it requires and the output labels that it produces.Lets now check the deployed models Metadata.Click on

83、 the CHURN_PRED name itselfOracle CloudWorld Copyright 2024,Oracle and/or its affiliates80Lets now check the deployed models REST endpoint Open API specification,by clicking its name under the URI.Oracle CloudWorld Copyright 2024,Oracle and/or its affiliates81The Open API specification contains the

84、details about how to access the model via REST,gives the server details,and specifies the expected inputs and the output labels that it produces,as well as available options.We are ready to deploy the Churn Model to an Application via REST1231.Back in the LiveLabs instructions,click View Login Info2

85、.Click Copy Value to copy the Database User Password3.Click Database Actions UI to go to the ADB Database Actions environmentIn the login screen that opens,type OMLUSER and the password AAbbcc123456When Database Actions open,we want to selectRelated ServicesOracle CloudWorld Copyright 2024,Oracle an

86、d/or its affiliates84Under Related Services you will find the Oracle Machine Learning RESTful services.Click open display the URLs.Oracle CloudWorld Copyright 2024,Oracle and/or its affiliates8512341.Click Related Services2.Click OML RESTful services3.Click Open4.URLs are displayed so user can copy

87、them for use in REST ApplicationsSelf-ServiceWatch the videos in beginning of each labGet help Self-ServiceFind the Need help?section in the sidebar of the workshopAsk us!I need help!Oracle CloudWorld Copyright 2024,Oracle and/or its affiliates86Wrap up and next stepsOracle CloudWorld Copyright 2024,Oracle and/or its affiliates87How to Implement Machine Learning Projects with Your Data and AutoMLThank YouMarcos Arancibia,Mark HornickOracle CloudWorld Copyright 2024,Oracle and/or its affiliates88