Get the Most Out of Oracle Analytics

In recent years, Oracle has introduced a wealth of powerful features in Oracle Analytics to enhance the user experience and reduce the need for IT involvement. However, most organizations only scratch the surface of their platform's potential and don't get the most out it. The modern Oracle Analytics platform can do much more than just export data to Excel. Features you can use include datasets, data blending, enhanced visualizations, predictive models, data flows and support for custom scripts.

Transcript

1. GET THE MOST FROM ORACLE ANALYTICS Federico Venturin | Oracle Analytics Architect
2. FEDERICO VENTURIN § Oracle Analytics Architect § Oracle ACE Associate ♠ § Oracle Analytics Ambassador § ITOUG Board Member § Based in Trebaseleghe Venice § fventurin.hashnode.dev § f.venturin@reply.it
3. INTRODUCTION
4. ORACLE ANALYTICS § Suite of tools and services for unified analytics across the enterprise § Supports all types of analytics use cases § Enterprise level reporting and analytics § Personal data exploration and visualization § Departmental data modeling and data science § Deployment options: § Oracle Analytics Cloud § Oracle Analytics Server § Oracle Analytics Desktop
5. ANALYTICS CLASSIC § Designed to create interactive, enterprise-level reports § Relies on: § The common enterprise information model (RPD) § Analyses to answer business questions § Dashboards to organize and share analyses § Agents to automate business processes
6. ANALYTICS PUBLISHER § Designed to create pixel-perfect, static report documents § IT builds data models § Business users build layouts using the browser or familiar desktop tools § High performance report generation and distribution engine
7. DATA VISUALIZATION § Self-service tool designed to make certain tasks easier for users, minimizing the need for IT involvement § Key features: § User-friendly, intuitive interface § Responsive design § Users can upload personal data § Data preparation capabilities § AI and ML capabilities
8. ORACLE ANALYTICS CLOUD § Cloud benefits: § Automatic updates and maintenance § Scalability and flexibility § Cost-efficiency § Security § Exclusive features: § Data replication § Oracle Analytics Day by Day § Natural language generation § No access to the underlying server and configuration files § Limits based on the licensed OCPUs (Plan Your Service)
9. ORACLE ANALYTIC SERVER § Full control over the platform § More customization options § Customers are responsible of maintenance and backup § New features are released only once a year
10. REFERENCES § Oracle Analytics Community § Oracle Analytics YouTube Channel § Oracle Analytics Cloud Documentation § Oracle Analytics Server Documentation
11. DATA VISUALIZATION FEATURES
12. DISCLAIMER § Data Visualization is not (yet) a replacement for Analytics Classic or Publisher § However, all Oracle Analytics users can benefit from Data Visualization capabilities
13. DATASETS § Self-service data models § Can contain multiple tables and join relationships between them § Files (XLSX, XLS, CSV and TXT) § Subject areas and analyses § Connections to supported sources § Quality Insights § Identify anomalies and outliers § Correct or replace values § 1-click data enrichments
14. DATASETS § Available as external subject areas in Analytics Classic and Publisher § Use cases: § 1-table data models § Combine multiple analyses § Combine multiple subject areas § Extend subject areas/analyses with personal data § They are not a replacement for a proper semantic model (RPD)
15. DATA BLENDING § Multiple datasets can be added to a workbook and combined using data blending § Dasets with attributes that share a common name and compatible data type are automatically matched § This matching can be customized
16. DATA BLENDING § Blending relationships are stored in the datasets and not in the workbook § They persist in Analytics Classic § Use cases: § Add facts to subject areas § Extend subject area dimensions
17. ENHANCED VISUALIZATIONS § Data Visualization comes with a rich set of visualization types § Visually impactful § Fully interactive § Responsive § They can be used in Classic dashboards and interact with dashboard prompts based on the same data source
18. MAP CAPABILITIES § Ready-to-use backgrounds § OpenStreetMap § Oracle BI § Oracle Maps § Ready-to-use map layers § USA Cities, Counties and States § World Cities, Continents, Countries, Regions, and States Provinces § Quick and simple association to data source columns
19. MAP CAPABILITIES § Ability to add backgrounds: § Google Maps § Baidu Maps § Web Map Service (WMS) § Tiled Web Map (XYZ)
20. MAP CAPABILITIES § Ability to add custom layers using GeoJSON files § The maximum upload size for an individual GeoJSON file is 100 MB (compressed) § The overall limit for GeoJSON files is 200 MB (compressed)
21. MAP CAPABILITIES § Ability to create custom map layers on uploaded images
22. PREDICTIVE MODELS § Oracle Analytics includes machine learning algorithms that enable users to train predictive models § Expertise in ML is not required § Use cases: § Predict values § Predict classes § Identify groups
23. PREDICTIVE MODELS § Use data flows to create, train, and apply predictive models § Oracle Analytics integrates with: § OCI Artificial Intelligence § OCI Data Science § OCI Functions § OCI Language § OCI Vision § Oracle Database § Machine Learning § Advanced Analytics § Oracle Autonomous Data Warehouse (e.g. AutoML)
24. SUPPORT FOR CUSTOM SCRIPTS § Extend machine learning and data curation capabilities of Oracle Analytics § Python/R scripts can be embedded in XML format and invoked from data flows § Must be enabled § Doc ID 2675894.1
25. ENTERPRISE ENHANCEMENTS
26. FINE-GRAINED DV PERMISSIONS § Oracle Analytics provides ootb application roles that include a fixed set of permissions to help you get started § Fine-grained permissions can be granted to (or revoked from) user-defined application roles
27. CONTENT MANAGEMENT § Allows administrators to view, manage, and change ownership of all content types available in Oracle Analytics, including datasets, workbooks, reports, and dashboards
28. ROLE-BASED FILTERS § Applied to datasets when accessed via workbooks or data flows by members of specific application roles § They do not work as row-level security rules in the repository § When there is at least one role- based filter, any user who does not have that role won’t be able to see any data in the dataset
29. RESTORE DELIVERY OPTIONS § Agents can be disabled or retain their status after a migration without manual handling or custom scripts
30. EMBEDDING ANALYTICS CONTENT § Oracle Analytics content can be embedded into applications, corporate portals or websites § No extra fees § Embedding methods: § iFrames § JavaScript embedding framework § Administrators must register the application’s domain as safe
31. EMBEDDING WITH IFRAMES § Users will be prompted to log into Oracle Analytics § Set up single sign-on or user federation to avoid this issue § Special characters in the URL need to be URL-encoded § Examples of URLs: § http://<ANALYTICS_URL>/analytics/saw.dll?PortalGo&Action=prompt& path=%2Fshared%2FSales%2FRevenue § http://<ANALYTICS_URL>/analytics/saw.dll?dashboard& PortalPath=%2Fshared%2FSales%2F_portal%2FSales%20Overview&page=Home § http://<ANALYTICS_URL>/ui/dv/home.jsp?pageid=visualAnalyzer&reportmode=full& reportpath=%2F%40Catalog%2Fshared%2FSales%2FTop%20Countries § http://<ANALYTICS_URL>/ui/dv/home.jsp?pageid=visualAnalyzer&reportmode=full& reportpath=%2F%40Catalog%2Fshared%2FSales%2FTop%20Countries&canvasname=canvas!2
32. JS EMBEDDING FRAMEWORK § Provides greater flexibility, but supports only DV content § Authentication methods: § Login prompt authentication § 3-Legged OAuth authentication § Token authentication § Use the Developer’s page to find the code that you need to embed DV content
33. OAC NOVEMBER 2024
34. AI ASSISTANT § AI-powered tool that helps users build visualizations using natural language prompts § Enabled on a dataset basis by indexing the dataset
35. AI ASSISTANT § Huge potential, but it is not a game changer § Requires a large OAC instance with 10+ OCPUs to work § Does not support subject areas § Can’t apply advanced analytics functions § Does not understand synonyms if they are not specified when the dataset is indexed § Can’t use existing calculated columns, or generate them on the fly § Can’t apply knowledge enrichments § Does not understand blending relationships § Can’t apply style formatting § Can’t apply conditional formatting § It offers functionality similar to the BI Ask feature but is considerably more expensive
36. GEOMETRY COLUMN SUPPORT § Allow a dataset to have a column of data type geometry for use in map layers and spatial calculations
37. CONDITIONALLY HIDE/SHOW VISUALS § Show or hide a visualization on a canvas based on the selected value of a parameter
38. CANVAS TEMPLATES § Create and share workbook canvas templates to enable workbook creators to quickly create visualization content
39. WORKBOOK THEMES § Create and share workbook and visualization styling properties, including color series and font settings to a workbook
40. PERFORMANCE TUNING MYTHS
41. MYTH 1 Add capacity to fix any Oracle Analytics performance issue
42. FACT § Add capacity alone is generally not sensible § Fix issues at root cause and you might offset the need for capacity at all
43. MYTH 2 Play around with configuration settings to fix any Oracle Analytics performance issue
44. FACT § Default values are generally good § Unnecessary fiddling without good reasons should be avoided
45. MYTH 3 Disable the query logging if you are having performance issues in Oracle Analytics
46. FACT § Query logging is needed to trace and diagnose performance issue § The overhead is neglectable when LOGLEVEL is less than or equal to 2
47. MYTH 4 Enable cache to fix any Oracle Analytics performance issue
48. FACT § Do not use cache as a mask for bad design § The actual problem is never addressed and will persist
49. MYTH 5 Enable Cache to allow for all necessary data transformations in Oracle Analytics
50. FACT § Data transformations should happen once at ETL time § Widespread usage at query time is indicative of suboptimal design, it’s difficult to maintain, and results in less efficient and complex SQL
51. MYTH 6 Oracle Analytics Datasets can effectively replace traditional Semantic Models
52. FACT § Datasets are not a replacement for the semantic model § Vertical Federation and Double Column features are not supported § Generate different (less efficient) physical queries
53. MYTH 7 Oracle Analytics Data Flows can effectively replace ETL tools and procedures
54. FACT § Data flows are not a replacement for ETL tools and procedures
55. MYTH 8 If a design pattern cuts development time, it must improve performance as well
56. FACT § Beware of overcrowded analyses, view prompts, and master-detail linking of views § Oracle Analytics retrieves results for all columns listed in the Criteria tab § View prompts and master-detail do not append any WHERE condition to the query § Create multiple smaller analyses, and remove excluded columns § Use dashboard prompts rather than view prompts § Use actions rather than sending master-detail events
57. MYTH 9 If a feature looks cool in Oracle Analytics, it will likely not impact performance
58. FACT § Beware of hierarchical columns, brushing, and embedded AI/ML capabilities § Hierarchical columns generate complex and long SQL § Brushing executes additional queries behind the scenes that may impact performances § AI/ML capabilities may require a lot of resources to load and manipulate data § Do not use more than 1 hierarchical column at a time § Switch brushing off by default in the System Settings page § Authors can enable it on demand at canvas level when required § Use AI/ML capabilities wisely
59. MYTH 10
60. FACT § Do not use Oracle Analytics as a feed for Excel-marts § Can you achieve the same within Oracle Analytics? § Exports to CSV should be used for data sets of > 100K rows § Do not use Excel formats for data sets larger than 100K rows § Exports of >500K rows should only use Publisher capabilities § Exports to CSV of 1M to 10M rows should be scheduled using Publisher § End users should not request large data set downloads on demand § Doc ID 1558070.1