Žiga Vaupot's Blog

I recently returned from Reading, UK, where I had the opportunity to attend the UKOUG 2023 conference . This year's event was particularly engaging, commemorating the 40th anniversary of the UK Oracle User Group. My participation at the conference had a dual purpose. Firstly, I attended to present, and while I was there, I had the chance to join several presentations that piqued my interest. I'm genuinely impressed by the innovative solutions showcased and the quality of work demonstrated by the presenters and their teams. My winners My standout presentation was LLMs are the Future of Conversational AI  by Antony Heljula of TPXimpact. Antony shared his experiences in developing chatbots that leverage large language models in conversational AI. Truly groundbreaking! A close second for me was Gianni Ceresa's (DATAlysis) presentation titled ID Please: Did You Already Ask That to Your Data?  Gianni delved into the crucial aspects of data lineage and governance, addressing the...

Using AutoML in Oracle Analytics Approximately one year ago, I have written a blog post Training and deploying AutoML models in Oracle Data Lakehouse . This blog post was part of my Oracle Data Lakehouse blog series, was focusing on training and deploying AutoML models in Oracle Autonomous Data Warehouse (ADW). Besides that, the post was also focusing on storing and managing data files in object storage and registering them as external tables in ADW. AutoML generated models are stored and deployed in ADW as any other OML models. If you want to use these models in Oracle Analytics you have to manually register OML model with Oracle Analytics and use it. There is a small detail regarding the location of data to be used with such a model - data has to reside in the same database where model is deployed. Not to mention, that user has to log into OML Notebooks in ADW and perform the training there. In the latest, March 2023 Update, release this is no longer needed as Oracle Analytics ...

Analytics and Data Oracle User Community ( AnDOUC ) Winter TechCast Days  took place between 15th and 17th Februrary 2022.  I am very honoured to make a small contribution at the  Machine Learning Day  where I presented my view on what business users should know about machine learning support in Oracle Analytics, presentation with the title  Business Users and Machine Learning in Oracle Analytics . Business Users and Machine Learning in Oracle Analytics Traditionally, the use of machine learning is in the domain of the data scientist. The latter has in-depth knowledge of machine learning methods and algorithms and strives for the most optimal preparation of machine learning models, which are then used for various analyses such as predictions, customer segmentation, anomaly detection, finding patterns and the like. Because the machine learning process usually takes a long time, even a few weeks, the data scientist is a rather “rare species”. Business users usuall...

Anyone who has done some machine learning modelling in Oracle Analytics, might have been missing a bit more of quality metrics to better assess the quality of trained models. By default any generated classification model would enable users to evaluate trained models by reviewing Quality metrics such as Precision , Recall , Accuracy , F1 and False Positive Rate . When presenting Machine Learning support within Oracle Analytics, the question about calculating  Gain and Lift are often asked. Until now, Oracle Analytics didn’t support this functionality and this has changed in the recent release of Oracle Analytics 6.3 . Lift and Gain Analysis In order to measure how the prediction model is better then not using a prediction model at all, Lift and Gain Analysis is used. Using other words, with Lift and Gain analysis we can find out what is the benefit of applying predictions based on a prediction model to the business. In Oracle Analytics, Lift and Gain Analysis is performed by vi...

Charlie Berger and Abi Giles-Haigh ran the hands-on lab session yesterday (11th May 2021) at Analytics and Data Oracle User Community's, AnDOUC ( https://analyticsanddatasummit.org/ ), TechCast Days, Spring Sessions with the title A Two-Step process for Detecting Fraud . I was invited to help with the HOLs alongside with Tim Vlamis and Edelweiss Kammermann .  The hands-on lab is about taking users through Oracle Machine Learning (SQL) Notebooks while using multiple use case scenarios for analysing insurance claims data and customer behavior data. Students were running the notebooks themselves to build and apply OML models inside Oracle Autonomous Database and then worked with Oracle APEX applications that leveraged OML’s insights and predictions ... So basically, what we did was: we took the Insurance Claims data set and ran 1-Class SVM algorithm, which is used for anomaly detection in order to identify suspicious claims.  For business (and any other) users we have an option ...