Tableau：2021年Tableau中的人工智能和機器學習研究報告（英文版）（19頁）.pdf

《Tableau：2021年Tableau中的人工智能和機器學習研究報告（英文版）（19頁）.pdf》由會員分享，可在線閱讀，更多相關《Tableau：2021年Tableau中的人工智能和機器學習研究報告（英文版）（19頁）.pdf（19頁珍藏版）》請在三個皮匠報告上搜索。

1、Artificial Intelligence and Machine Learning in TableauPrepared by Joshua Gillmore and David SearsContentsVisually driven AI and ML.3AI throughout the analytics flow.4Explain Data.5Ask Data.7Analytics Pane models.9Predictive analytics in calculated fields.11Einstein Discovery integration.13Analytics

2、 integration.16Extensions.18Learn more.19About Tableau.19Visually driven AI and ML At the center of the Tableau experience is seeing and understanding data.Combining Tableau visual analytics with the power of Artificial Intelligence(AI)and Machine Learning(ML)keeps the human at the center of decisio

3、n-making while unlocking advanced analytics across the entire organization that is typically limited to highly technical teams within an organization.The platforms ease of use provides users at all levels of expertise the ability to tap into visually-driven,advanced AI and ML-based analytics to get

4、immediate time to value and uncover unanticipated insights.At the core of Tableaus approach to AI and ML are three themes:1.AI and ML naturally integrate into the users workflow to enhance their analysis while avoiding disrupting their thought process and flow.2.Full-featured AI and ML capabilities

5、are usable regardless of experience and expertise.3.AI and ML results are transparent,interpretable,and trusted.4Artificial Intelligence and Machine Learning in TableauAI throughout the analytics flow Tableau offers a complete,integrated analytics platform.AI and ML will enhance decisions throughout

6、 the analytics flow.Whether its handling missing data or understanding the why behind a data point,Tableau has integrated central AI and ML functions to assist human-decision making throughout the analytics flow:1.Explain Data Utilize intelligent AI modeling at the click of a button to quickly under

7、stand the Why behind a data point in your Tableau visualization.2.Ask Data Create data visualizations and dashboards using simple questions in natural language.3.Tableau Analytics Pane Drag and drop machine learning to predict future outcomes based on historical data with forecasting,determine futur

8、e trends for your data using various models,or understand the relationships between data points with clustering.4.Integrated Predictive Modeling Functions Engage in predictive analytics and fill in missing data using Tableaus built-in calculated fields.5.Einstein Discovery in Tableau Build AI-powere

9、d predictive and prescriptive analytics with automated&guided model building,and embed these actionable custom predictions anywhere users can see Tableau.6.Supported R,Python,and MATLAB Integrations Make the most widely used data science languages interactive in Tableau with TabPy,Rserve,and the Ana

10、lytics Extensions API.7.Tableau AI/ML Extension Ecosystem Use drag and drop extensions to take the power of AI and ML Tableau Partner technologies through Tableaus Extension Gallery.5Artificial Intelligence and Machine Learning in TableauExplain DataExplain Data accelerates analytics by leveraging A

11、Is power to explain exceptional data points and their drivers in visualizations with a single click.Based on advanced statistical modelling,Explain Data presents a focused set of transparent explanations to answer the why and avoid spending time on dead ends.Use:Identify potential policy levers in y

12、our data for future exploration.Determine the standout programmatic funding and distribution for a given postcode at the click of a button.This series of answers means a policymaker can quickly determine whether funding meets a geographic areas needs.Accessing Explain Data is a tool-tip away for all

13、 of the data points.This ease of access enables all users to quickly access augmented intelligence explanations when they find a data point of interest.No additional input is required as results generate automatically based on Explain Datas selection of appropriate data points.6Artificial Intelligen

14、ce and Machine Learning in TableauExplain Data will intelligently select the underlying data points to compare and draw out avenues for the user to explore the“why”behind their data.Explanations are visualization centric to quickly show the user whats unique about Explain Datas findings.For the more

15、 advanced users,selecting the Explain Data fields under your data in the menu will let them refine the model and experiment with various combinations on the fly.The Explanations Window and pre-built worksheets use plain language to explain the data points uniqueness and underlying relationships with

16、 other data.Dive deeper into the insights and transparent explanations through automated visualizations in the worksheet.7Artificial Intelligence and Machine Learning in TableauAsk DataWith Ask Data,you can use natural language to ask questions and instantly get a response right in Tableau without u

17、nderstanding the datas structure.Answers come in the form of rich data visualizations without understanding the data structure to get answers faster.Fully integrated into the Tableau platform,Ask Data works with existing data sources with one additional step.Use:Ask Data will democratize any data so

18、urce with a single step.Evidence locked up in data now becomes accessible to knowledge workers across the organization by querying the data through simple conversational language.This means no drawn-out implementation process to create a dashboard.Instead,any data source can become the foundation fo

19、r evidence-based policy.8Artificial Intelligence and Machine Learning in TableauAccessing Ask Data is available by selecting any data source within your Tableau Server.Once in,Ask Data presents a free text query box found throughout popular search engines.When typing into this intuitive interface us

20、ing plain language,the user is then presented with a series of options that Ask Data thinks will answer their question.Getting at the heart of the answer will usually require refinement,so Ask Data allows for follow-up questions to delve deeper and provide the recommended visualization containing th

21、e answer or enable the user to select their preferred option manually.While typing queries,Ask Data will show relevant values in the data,their fields,and suggest functions,such as totals,to produce a visualization.Once the visualization is generated,visualizations will contain a list of underlying

22、questions to retrace the thought process behind the visualization.Ask Data results are presented directly through text or intelligently selected visualizations without a need for prompting.For users unfamiliar with the data or creating visualizations,Ask Data will automatically create a visualizatio

23、n to help answer their questions without requiring additional expertise.Advanced users can use Ask Data to create a new dashboard,work through complex and large datasets by asking probing questions that may need to find a known value in a column,or answer questions on-the-fly during meetings.9Artifi

24、cial Intelligence and Machine Learning in TableauAnalytics Pane modelsThe Analytics Pane is Tableaus equivalent of having a drag and drop machine learning-enabled statistician on demand.In addition to the typical statistical workhorses commonly used,like mean and median,the Analytics Pane offers a s

25、eries of models to help predict your datas likely future with forecasting,its overall patterns with trends,and its closest relationships with clustering.Use:Understand the likely direction of organization expenditures using machine learning-based forecasts and trends.Like a rainy day,use this inform

26、ation to better prepare.The Analytics Pane is available throughout the visualization creation process in Tableau Desktop to immediately enrich data through drag and drop functionality.Functions will provide guidelines on the required data to visualize and produce a series of options with visual queu

27、es to help users of all levels make the right decision and understand how their data will impact the outcome.This drag and drop functionality makes it easy to experiment and determine the best fit for data-driven decision-making.Further interactions with the data through visualization will dynamical

28、ly update the analytics outcomes to continue assisting human-driven decisions.10Artificial Intelligence and Machine Learning in TableauTableau is using machine learning intelligence to select the best fit models based on available data.Users of all levels applying these models will benefit from Tabl

29、eaus intelligent guidance for the best fit.Additional visual cues throughout the drag and drop process make it intuitive and easy to understand what data the analytics is considering and the type of outcome a user expects.This simple method to apply models means experimentation for users to find the

30、 best suit quickly.Once the calculation is complete,the calculated field becomes a part of the drag and drop selection for inclusion in dashboards for visual exploration.Tableau automatically produces visual confidence bands where applicable or allows users to generate model descriptions that will p

31、rovide in-depth information on underlying information that may include descriptive statistics,formulas,or variables,to name a few.11Artificial Intelligence and Machine Learning in TableauPredictive analytics in calculated fieldsTableaus Calculated Fields produce data on the fly using intuitive funct

32、ions akin to those found in widely-used spreadsheet software.Predictive analytics functions are available within the list of functions to help fill in missing data and understand how changes in data are likely to affect the outcomes in other places.Generating these new calculated fields means they b

33、ecome available for users to integrate into their visualization or for analytics consumers to see the results as they explore visualizations.Use:A perfect world contains perfect data.Real-world scenarios generally have missing and incomplete data.Predictive modeling can fill these holes in your data

34、 using previous values or other sources to ensure continuity and fewer assumptions in your decision-making process.Creating a calculated field is as simple as right-clicking on the Data Pane and selecting the option.A series of functions are available for the analyst to choose that include the two p

35、redictive functions.Entering these functions will prompt the user to specify the required dimensions or measure,in addition to validating to ensure all the required data is present.12Artificial Intelligence and Machine Learning in TableauAnalysts and data scientists are the users who will intuitivel

36、y understand how to create predictive outcomes that will benefit key business questions.Once created,these calculated fields are available for use by the broader group of business users when making Tableau visualizations and analytics consumers exploring visualizations with predictive embedded.Any v

37、isualization containing the predictive calculation will appear in a pill where a user can open up the calculated field to understand further the fields feeding the predictive calculation and any comments originally included.13Artificial Intelligence and Machine Learning in TableauEinstein Discovery

38、integrationWith Einsteins powerful AI technology,you can quickly operationalize predictive intelligence for your entire organization in Tableau without writing a single line of code.Einstein Discovery automatically examines millions of data points from many sources,enabling analysts to rapidly train

39、 and deploy predictive models.Tableau surfaces predictions,allows for visual interactivity to explore possible outcomes,and includes the underlying reasons for the predictions.Use:Einstein Discovery will help process a pile of grant applications and reduce total time on the evaluation.By taking your

40、 previous grant applications and their outcomes,Einstein will provide predictive insights that will help identify high-quality referrals and provide explanatory insights when you need to shift your funding portfolio and assess additional risk.14Artificial Intelligence and Machine Learning in Tableau

41、Einstein Discovery in Tableau integrations benefit business users to seasoned data scientists.The new dashboard extension arms Tableau users with automated discovery,no-code machine learning that builds automated predictions,bias protection,and transparent predictive analytics within consumable Tabl

42、eau visualizations for business users.Einstein Discovery calculated fields allow seasoned analysts to generate new data using Einstein Discovery to enable users with a drag and drop feature into Tableau visualizations for real-time predictions.The Tableau Prep upstream predictive data enrichment mea

43、ns that Tableau users of all levels downstream benefit from advanced analytics and predictive factors towards better decision-making.15Artificial Intelligence and Machine Learning in TableauEinstein Discovery provides predictions in a consolidated view that identified top predictors,the relationship

44、 strength to the outcome,and the strength.Going deeper into Einstein will reveal additional findings and details on the models supporting the outcome.Integration within Tableau will display the raw data underpinning the predictions by record for visualization and sharing.16Artificial Intelligence an

45、d Machine Learning in TableauAnalytics integrationWhen you convert your Python,R or Matlab scripts to run in Tableau,you open up a world of new opportunities and democratize data science by making advanced algorithms accessible to everyone in your organization with a simple drag and drop.It also all

46、ows your teams to see,understand,and explore a scripts results in a visual,interactive format.Use:Deep learning and natural language processing is typically tricky for an analyst,subject matter experts,and decision-makers to interpret.Pairing the underpinning technologies with data visualization in

47、Tableau means opening up these technologies for programmatic and policy decisions in a consumable format that minimizes the data science teams overhead and allows them to focus on their expertise,data science.Even users without any scripting background can connect scripts for enrichment upstream in

48、Tableau Prep,or use the visual queues in Tableau Desktop.This means that complex scripts created by data scientists are available for use within the Tableau analytics process.Re-use of content previously created in future dashboards is also an option to take advanced analytics in new directions.Tabl

49、eau offers more advanced solutions for more experienced users by combining data generated on the fly through calculated fields or tapping into stored predictive models using TabPy.Tableau Prep and Desktop include multiple locations to integrate Python and R scripts.Connecting to TabPy or R services

50、is a simple process that can be started in as little as two values.Implementing a script is a guided process in the Tableau Prep user interface.It uses Tableaus built-in calculated fields in Tableau Desktop by including the script directly in the prompt.17Artificial Intelligence and Machine Learning

51、 in TableauRserve and TabPy consoles and logs provide real-time feedback on the functions being performed and issues encountered.Throughout the process,users can open up the R or Python functions being applied in the calculated fields or use Tableau Prep to access the scripts location for direct acc

52、ess with the proper permissions.18Artificial Intelligence and Machine Learning in TableauExtensionsTableau Extensions are widely available through the Tableau Extension Gallery and maintain a simple drag and drop approach for implementation.Tableau has a rich ecosystem of partners utilizing the Tabl

53、eau Extension ecosystem to enhance and augment Tableaus data Visualization capabilities that include extensions with AI and ML capabilities.Use:An AI and ML technological revolution is abounding.Several of these technologies exist across many Tableau partners.Extensions allow for simple integration

54、to realize all the benefits of these technologies in your Tableau analytics workflow.This means plain language interpretation of your results in Tableau by combining the natural language generation technologies.Newly created enriched data in these extensions may assist with predictive outcomes,scena

55、rio analysis,and what-if analysis.Combining this enriched data with Tableau visualizations will allow for real-time exploration of results that update on-the-fly through Tableau user interaction.In some instances,the extension may include new data visualizations natively built on the AI and ML resul

56、ts to assist with the analytics process for all experience levels.Extensions will typically contain simple graphical user interfaces for any level of Tableau user to simplify enrichment of their data using AI and ML technologies with no programming required.19Artificial Intelligence and Machine Lear

57、ning in TableauLearn moreTableaus latest releasesGovernment AnalyticsFederal Civilian AnalyticsDefense Intelligence AnalyticsState and Local AnalyticsAbout Tableau Organizations around the world are using Tableau to share data and insights and keep their employees and citizenscitizend engaged.Find out how our platform can help your organization to see and understand their data by enabling self-service analytics,allowing collaboration,and swift insight-to-action.Try Tableau for free today.