Posts

Showing posts from 2020

Oracle Analytics: Tips & Tricks - Webinar from MakeITWeek2020 (SIOUG) conference

Image
Together with my colleague Mojca Gros of Qubix, we have delivered webinar session with the title Oracle Analytics: Tips & Trick at this year's SIOUG annual conference MakeITWeek2020. In a record time, Oracle Analytics has undergone a transformation from a “old-fashioned” business intelligence tool to a modern, mature, and comprehensive analytics platform that runs in the cloud or, if you want, on a user’s server. New functionalities are being introduced on a very frequent basis, and these are very difficult to keep track of. Be it in the field of data visualisation, data preparation for analysis, machine learning or infrastructure changes. With the development of projects, we clearly face many challenges, which also affect how to plan business analytics solutions.  Webinar discusses some of the challenges and some Tips & Trick that are now available. We are talking about and demonstrating: data discovery and explainability data preparation and enrichment data visualisation

Market Basket Analysis with Oracle Machine Learning and Oracle Analytics

Image
Introduction Recommendations are usually always there to stimulate you to buy additional products or services regardless if you are shopping online or in a store. Sometimes these recommendations are just intuitions of sales manager, but more often these are prepared using some machine learning algorithm, for example Association Rules. When Association Rules modeling is applied to transaction sales data, the model is called Market-Basket Analysis. Market-Basket Analysis can be used in various situations, from direct marketing, sales promotions, discovering business trends, but also for effectively managing store layouts, catalog design or for exposing cross-sell opportunities in a web store. Since we I am not going to discuss theory about Market Basket Analysis and Association Rules, please find more information here:  https://docs.oracle.com/en/database/oracle/oracle-database/19/dmcon/data-mining-basics.html, https://docs.oracle.com/en/database/oracle/oracle-database/19/dmcon/associati

Having issue with opening RPD file after you've downloaded a snapshot?

Image
... well, I certainly did. As usual I was taking the shortcut. Once snapshot was created, I downloaded the .BAR file, renamed it to .ZIP, extracted it, found the .RPD file in extracted folder structure and ultimately opened it. It worked for me so far, but I learnt my lesson a couple of days ago. I don't know, maybe I wasn't paying attention, but the case was: "Invalid password". Basically I found a solution quick at Oracle Support under this very link , however, let's take a look at the whole process. When snapshot is created, one would usually download the snapshot in the form of the .BAR file. After the snapshot password is provided, download can begin. In my old habit I then simply renamed .BAR file into .ZIP file which is then extracted. In the extracted folder, .RPD file is located in the \datamodel\rpd subfolder (in the screen below, I copied default.rpd file to Desktop folder).  And if you try to open the .RPD file this way from Model Administration Tool (

Using Oracle ADW machine learning models in Oracle Analytics

Image
Oracle supports machine learning within several tools and technologies. One of the oldest technologies that supported machine learning algorithms in database is Oracle Data Mining. This has become the core for Machine Learning for SQL which is part of Oracle Database (at no extra cost to Oracle Database users).  Oracle Analytics now supports integration with Oracle Machine Learning for SQL and can use ML models that are stored in Oracle ADW. In my blog posts I am exploring options which are available to users by using rich and powerful Oracle Autonomous Data Warehouse support for machine learning and flexible data management and visualisation of Oracle Analytics. Using Oracle ADW machine learning models in Oracle Analytics ( part 1 ) and Using Oracle ADW machine learning models in Oracle Analytics ( part 2 ) 

Using Oracle ADW machine learning models in Oracle Analytics (part 2)

Image
In  the 1st part of this blog we have discussed how to create a regression model for numeric prediction of housing prices. I have posted similar post Housing Price Prediction in Oracle Data Visualization , describing how to predict house pricing with machine learning algorithms and data flows available in Oracle Analytics. Register ML model from ADW in Oracle Analytics  Let’s switch to Oracle Analytics now. We need release 5.7 and higher to register and use ML model from ADW. The first step which we need to do is to register ML model from ADW with Oracle Analytics.  If there is no database connection, setup this first from the Create Connection menu from Home Page. Navigate to Machine Learning and then from actions menu (top right corner) select Register ML Model: If no connection is available, choose Create Connection. We have created our connection in the step before, therefore the list of connections is displayed. Choose the connection ... ... and then choose ML model to register.

Using Oracle ADW machine learning models in Oracle Analytics (part 1)

Image
A few weeks ago I was exploring how can Oracle Analytics can be used for housing prices prediction. In my post  Housing Price Prediction in Oracle Data Visualization  I am discussing the tools available in Oracle Analytics to perform the task. In this post (actually there ware 2 parts) I am looking at options that we have in Oracle Analytics if we wanted to use machine learning models that reside in Oracle Autonomous Data Warehouse.  Oracle Data Mining was an Oracle Database option for more than 2 decades now. With advancement of machine learning support in database, it became part of Oracle Advanced Analytics database option. Developers were able to use Oracle Data Mining by using Oracle Data Miner which was part of the Oracle SQL Developer.  Oracle Data Mining is PL/SQL based set of libraries which supported several supervised and unsupervised data mining functions such as classification, regression, clustering, time series analysis and other functions. These functions were support b