Predictive Science in Not-for-Profit OrganisationsGreg RobertsJuly 8th 2019
Machine Learning, AI, Data Science, Deep Learning; there’s a lot of jargon out there, and equally as much FOMO. In this blog, I discuss our approach to projects in this area, why it’s not as outlandish as it might seem, and how NFP organisations can start leveraging this area to generate incremental returns, quickly, and with minimal risk.
At our incremental transformation event for charities, I (Greg Roberts, Data Scientist- cloudThing), delivered a session on what Predictive Science is and how it can be applied in the Not for Profit sector. If you weren’t able to make the event, I’ve decided to write a blog on the key points so that you can explore this area further – I hope it’s of value to you! So, onto the blog...
Let’s start at the beginning and get some background on the matter: what is Predictive Science and why is it called ‘Predictive Science’? There are many buzzwords around data services in IT at the moment, such as Artificial Intelligence, Machine Learning, Data Science, Predictive Analytics, Big Data - everyone seems to be saying that you need these in order to keep up with the competition. It can be quite overwhelming trying to define where to start, and how to separate facts from the hype.
All of these buzzwords are centred around one general concept: using statistics and algorithms to make decisions and predictions based on data, automatically. The volume of data being generated has increased tremendously, and the number of products and frameworks to create value from data have matured with it. This means that now is probably the best opportunity and time to see how you can leverage your data in order to gain an insight into your customers or your organisation, through the data you hold. The technology can only get you so far - you need a combination of human insight and software to turn old spreadsheets and databases into something worthwhile.
The principle is the same across all projects, from Prescriptive Analytics (using descriptive statistics to understand data in an interactive way- so for charities, you could begin to analyse the data in relation to your donors; when they joined, who has left, is there a trend of increase/decrease in donors, and then begin to explore why they left etc), right through to cognitive services (like Alexa, or helpful chatbots on your organisation’s page where you can solve problems in many different areas (giving donations, memberships etc). This is why we like to think about the whole area as ‘Predictive Science’ and from these few examples, hopefully you can see the benefit and positive impact that these aspects can bring.
It’s important to think of Predictive Science solutions in the context of augmenting human ingenuity rather than replacing it. Now there’s some food for thought. Technology these days is created to help complete tasks more efficiently and for Not-for-Profit organisations, this could prove beneficial. By automating certain tasks, more time will be able to be spent on jobs such as event planning, acting upon supporter feedback, and much more. To keep it realistic, even the best algorithms will only give you the correct answer 85-90% of the time- yes that’s still quite high but it’s not 100%. An example can be based on chatbots; when an enquiry is too complicated for the algorithm to suggest an appropriate answer, it will be passed over to a human to deal with so that the correct outcome can be achieved.
During this session, I demonstrated many different solutions from Microsoft Azure and the three that I want to focus on today are Azure Notebooks, the link into Dynamics and Power BI, and the cognitive services:
The biggest change to this domain over recent years has been the proliferation of SaaS companies offering pre-built, end-to-end solutions for specific sectors, with Predictive Science at the heart of their offering. This provides a new dilemma for organisations wanting to make use of this technology. Do you buy the solution pre-made from a vendor, or do you build your own solution from scratch?
There are pros and cons to either approach. Buying can significantly reduce the time and effort required to generate ROI but may involve a massive up-front investment, which could make it a much harder sell to stakeholders. On the other hand, building in-house will mean your organisation has full ownership of all the insights generated through the development process (rather than these merely enriching your vendor’s offering for other clients!) but resourcing such a project could also involve significant expense.
The third option would be to get in touch with cloudThing as a partner for a project. We have deep experience in the field, and we know the frameworks, techniques and tooling needed to mitigate the risks and start delivering value early. With us, you will get all the benefits of building (owning all the infrastructure and insights), plus all the benefits of buying (bespoke, best in class software), without the negatives of either option.
Another option for building if you don’t have all the staff required, could be to use our Team as a Service, where we would ‘give’ you the staff and expertise that you need in order to create the bespoke tooling to fit your organisation’s needs. Find out more on our website, https://cloudthing.com/method/ct-method-taas.
So, to conclude this blog, we need to remember five simple things:
If this blog has helped you understand Predictive Science further and open your eyes to its opportunities, why not get in touch with us today for a chat to see how and if we can help your organisation reach its full potential.