APAC Data Analytics Day (Online)

Login

Machine learning technologies are quickly transforming almost every sphere of our lives. From how we communicate to the means we use for transportation, we seem to be getting increasingly addicted to them. With these rapid advancements, massive amounts of talent and resources are dedicated to accelerating the growth of the technologies. Implementing open-source machine learning stack gives the organization more eyes for bug fixes and a wider set of features to cater to their machine learning requirements but at the same time organizations are tackling data security and access control regulations. While Open-source stack is preferred in many large organizations around the world, there is some serious decision making associated with it.

Machine learning models are being increasingly used to make decisions that affect people’s lives. To paraphrase Uncle Ben from the Spiderman comics (even though I am a DC fan) “with great power comes great responsibility”, this is true as the models that we build can adversely or favourly affect demography’s of a population. So, to ensure that the model predictions are fair and not discriminating we need to ensure our models do not contain an inherent bias by those who design them.
In my presentation I look to explore the idea of inherent model bias and potential strategems to tackle this issue.

With a mission to become truly data driven, companies make significant investment in establishing data analytics platforms but many of them fail to realise the full potential due to poorly executed strategy underpinned by ineffective organisation design. In order to embed data analytics culture at the heart of the organisation, the importance of fit-for-purpose organisational model cannot be overlooked. How do you source, retain and deploy talent? Should you go with centralised, decentralised or hybrid model? How do you build a COE to that can maximise knowledge sharing, yet operate effectively to deliver value?
This talk helps you explore what type of model would be most effective for your organisation and how best to implement them.

Unlike typical software, Machine Learning applications do not have straightforward rules and usually act as a black box. Lack of explainability and interpretability reduces the trust of the product owners, legal counsels, and non-technical end-users. This talk helps you to understand the importance of interpretability and the trade-offs between interpretability and model accuracy among various machine learning models. It gives you some ideas to make machine learning more transparent and less of a black box.

Lunch Break

The Data Fabric’s automated data engineering provides the business with the mechanisms to manage sparse data sources so it can create an integrated information and analytics Virtual Data Platform, and agile business self-service while guaranteeing 100% data quality.
For all intent and purpose, the ‘Virtual Data’ Platform (Data Fabric) refers to the set of data objects, including business entity types and business rules, that represent a Conceptual ‘Target’ Business Data Model (schema), which exist in memory within the Data Fabric Platform, and may or may not be persisted in a physical repository or a Virtual Data Warehouse (VDW). Once the Single View process applies all data integrity and data quality rules, the data objects are available to be accessed by the end-users and consumer applications directly from the Data Fabric Platform. This presentation explains a process that allows for agile business self-service to exploit business intelligence tools such as Qlik Sense, PowerBI, Tableau, MicroStrategy, etc., while providing for data provenance and data lineage as it relates to self-service.

How to identify organisational needs. How to find the right candidate pool. Budgeting correctly for staff. Personalising training so it WORKS for everyone. Empowering and trusting in staff. Making sure everyone is a committed team member and has "skin in the game". (Important for retention). Various other facets of retaining staff, and building effective teams.Data Analytics Talent: Recruiting, Hiring, Organizing, Training and Retaining

Many big data and advanced analytics investments fail to deliver significant and lasting benefits. Technology is rarely the issue. O’Reilly recently delivered the results of their study on AI adoption in the enterprise in 2020. Culture, skills gap and data quality as the as the biggest obstacles to businesses leveraging value from their data.

Most large companies are incredibly well-placed to generate significant competitive advantage from their data. They’re sitting on a wealth of data; they have talented analysts and scientists, shelfware galore and plenty of pain points to work on. So why aren’t they generating success?

AGL has embarked on an aggressive data and analytics transformation and is managing the significant change through four levers:
• Mindset – shifting the collective mindset of the entire organisation to embrace data and analytics in decision making focusing on building executive sponsorship, re-aligning strategic imperative and developing new organisational metrics
• Skillset – uplifting capabilities in technical expertise, data literacy and story-telling, and data driven leadership across the organisation through a purpose-built learning platform, guilds and career mapping
• Toolset – demonstrating fast value through ‘activation’ of use-cases and initiatives requires a new model of engagement between technology and the business; this lever is the most difficult, but wins over the most sceptical of executives and is critical to success if it is planned and not left to chance
• Dataset – embedding pragmatic data governance and stewardship that balances data privacy and security while not tying up data in red tape and bureaucracy.

I’ve led the change in similar programs at Suncorp, NAB and now AGL. I can talk to why and how to do this and what the results are.

- Building a winning culture
- Stimulating the collaborative culture with enough opportunities to socialize and strengthen relationships
- Getting away from Top Down hierarchical management approach to more equal footing accountability model.
- Everybody takes ownership of the work and knows how that will contribute to the bigger team success.
- Working style is more entrepreneurial and has continuous improvement.