Big Data LDN meetup

Published by Clare Patterson on

We had a double bill of presentations at the London Project Data Analytics meetup on 13 November covering benchmarking and establishing a data science team.  

Tim Podesta, Independent Consultant 

Tim provided an expert practitioner view on benchmarking, sharing his experience of 35 years of working within BP, including leading their benchmarking capability and working closely with the Construction Industry Institute.  

He discussed the key themes of front end planning, team alignment and complexity, which emerged from years of work collaborating closely with the construction industry institute.  

The team alignment aspect looked at leadership, communication, effectiveness of meetings. Note: Its worth taking a look at Kelvin McGrath’s work on this, which leverages data science to extract insights and identify lead indicators which wouldn’t have previously been possible.  

Tim also commented on the correlation between schedule slippage, project complexity and team cohesion. He highlighted that this work was based on sampling the data rather than streaming data analytics. He saw the potential for advanced analytics to drive new insights in this area in real time.  

Tim also drew our attention to a recent survey of industry project benchmarking practitioners, where 80% expressed a dissatisfaction with their approach. They main reasons for this were that they: 

  • Didn’t have the data people  
  • Didn’t have the data processes  
  • Didn’t have the data infrastructure  

Thanks for sharing some great insights Tim. We appreciate it. 

Note: I’m in the process of writing a paper with Tim on next generation benchmarking, exploring how we leverage streaming project data rather than sampling data. We hope to get it completed soon.  

Tim’s slides can be found here Benchmarking Data Analytics Meet-up Nov 14th

Dr Jan Teichman, Head of data science at Zoopla provided a very engaging presentation on how to set up a data science team and position it for success.  

He shared some interesting facts:  

  • 3/4 of data science projects are collecting dust.  
  • 85% of Big Data projects don’t move past the preliminary stages 
  • 80% of analytics insights will not deliver business outcomes through 2022 
  • 80% of AI projects will remain alchemy run by wizards through 2020 

He highlighted a number of challenges with establishing a developing a data science team, specifically: 

  • Motivation. Is there senior level engagement. Are they truly engaged and invested in the project or is a vanity project, or a pet project. Does the senior leadership team really understand the opportunities and challenges involved, including how the outputs of the project will be utilised.  
  • Preparation. How solid are the foundations for the project. Is the data available and is it aligned to the problem statements that the organisation is seeking to resolve. Is the team structure appropriate to the challenge that the team is trying to resolve?  
  • Hiring. Is the organisation clear on what skills it needs, from data engineering, data analysis through to data science. Which ponds are the recruiters fishing in and do they have a compelling case to recruit the very best candidates or are there sexier jobs in other professions?  
  • Delivery. Is there a defined process to take the solutions from concept through to production? Or are the solutions on laptops and never see the light of day?  
  • Retention. Data scientists are in demand. Do you have the opportunities to ignite their interest? Will it stretch them and give them a sense of purpose. How will their story compare to their peers when they are in the pub? Far too often, solutions do not reach production and the sense of personal fulfilment degrades.   
  • Jan really challenged the concept of Scrum. He said that although it helps to manage team activity and velocity, it didn’t address motivation, engagement, retention or actual deployment into production. He suggested a different and more holistic approach. I saw a lot of merit in his arguments.  

He was a big advocate of a product based approach, differentiating between proof of concept and proof of value. By taking a product based approach there is always a clear linkage back to the business problem that data science is trying to resolve.  

He was an advocate of ensuring that the project moves the dial and delivers a change against a set of agreed metrics. But productionisation is not enough. There is a need to ensure that the production models remain valid, particularly where the model may drive a change in behaviour.  

Jan’s slides can be found here Data Science Delivery – Beating The Odds (Circulation)

Martin Paver is CEO/Founder of Projecting Success. In Dec 2017 he saw the opportunity to transform how projects could be delivered through the application of advanced data analytics and founded the London Project Data Analytics meetup, which has expanded throughout the UK and has grown to a community of ~4000 people. He delivers strategic and tactical solutions that integrate project management and leading edge data science and analytics.

Categories: Blog Post

0 Comments

Leave a Reply

Your email address will not be published. Required fields are marked *