What Is Data Science And How Can It Help Local Authorities? | Peak Indicators
To understand how local authorities turn data into key insights that improve public services we must first define and understand the field of data science. We’re in the middle of a revolution, a data revolution, in the way data is shaping knowledge discovery, driving decision making and impacting on society.
Download The Full Public Sector Playbook Here
An enormous amount of facts are now being collected. We have access to far more data than ever before, coming in multiple forms and from a multitude of sources — there’s not just numerical data but vast amounts of text and multimedia data, as well as real-time streaming data from IoT devices, social networks and much more besides.
Data have become the raw material feeding data-driven decision making and AI systems in government and business. However, while many organisations collect a lot of data, most are not putting it to work in an effective way. Those that do can gain deeper insights, help people make more informed decisions and drive automation. Those that do are using data science.
Defining Data Science
But what exactly is data science? ? It’s not an easy concept to define, but in short it involves principled and scientific approaches to collecting data, building models, finding key insights and using data to answer questions and guide decisions. It’s science — the endeavour to discover the unknown — using data. It’s about using computers to find hidden patterns, connections or relationships in the data that can be used for inference or prediction. Along with Data Analytics, AI and Data Engineering, it’s shaping and transforming the way many enterprises are working.
But wait a minute, isn’t this what existing fields like statistics and data mining have been doing for years? Is there actually something new about data science or is it just hype? Well perhaps what defines data science today is that within the current data revolution we need new ways of applying existing methods — or even to develop new ones — to extract value from data. And to do this we’ll borrow techniques from many different fields, including maths, statistics, AI, data visualisation and computing.
So what can we use data science for? Your goals could be many:
- Discovery — for instance, finding that a very high proportion of students who regularly miss school live in a certain area
- Decision making — such as, targeting families in that area with your inclusion work to drive more regular attendance
- Prediction — for example, working out what levels of traffic an area is likely to face in the coming years
- Optimisation — for instance, using limited resources in the most cost effective way
By applying statistical and computational methods (algorithms and machine learning), we can automate the process of knowledge discovery, helping people identify and understand things happening in the real-world and empowering decision makers. But it’s even more than that. Data science is also being used to build data products, automate and improve processes, and even generate new business models.
Download The Full Public Sector Playbook Here
How A Data Science Project Works
Although data science projects can come in all shapes and sizes, there are some common stages. The process is often captured by standard methodologies, such as the CRISP-DM, OSEMN or TDSP frameworks. But they all tend to share similar stages.
Let’s walk through the steps of a typical data science project.
It starts by defining a problem to address with data. This may be driven by a question or business need and will often require talking with stakeholders. It’s important to have a goal for the project from the outset, although this may evolve throughout the project lifetime.
Next you need to identify and gather appropriate data that can be used to answer the question or address the problem at hand. This stage may involve collecting data from multiple sources and require detective work to understand what you have.
After that, you need to prepare and clean up the data because in practice data is messy. This is a key step, since poor quality data produces poor predictions and recommendations. Cleaning may involve correcting missing or duplicate values, combining conflicting data sets and removing outliers.
As you prepare and clean the data you’ll often need to go through a process of exploratory data analysis. This ‘playing with data’ is actually really important and may involve descriptive and inferential statistics, along with data visualisation, to summarise and understand the data.
Once you understand the data, you can begin building models to make inferences and predictions. This will often utilise statistical methods or AI and Machine Learning algorithms. This is also another scientific process where you will use an experimental design to compare and evaluate methods and develop valid models.
In practice, after experimenting and developing models you will have to operationalise and deploy them into existing systems and processes. This often requires data engineering and software development expertise.
In addition, people are often trying to use insights to drive decision making, so communication is also key. It’s about reporting the findings, and using visualisation and storytelling to help people understand them. This is an important step as you can have the best models and insights in the world, but unless they are communicated and used they will have no real-world impact and not lead to change.
This is how a data science project usually works, but it isn’t usually just a one-shot process. Projects could go through these stages once or twice or more, making adjustments to develop deeper insights and better models to create better and better predictions and recommendations.
Data Science In Local Government
There are a range of pressures facing local authorities. On the one hand, we’ve seen a decrease in government funding, cuts to services and staff counts, and less time or resources available.
On the other hand, local governments need to work with new sources of citizen data (for example social media), to tackle problems that are only increasing in number and complexity.
Much local government work involves deciding when and where to supply services and make interventions. These decisions are often complex, and a lot of the work tends to be reactive.
This has led to local authorities making use of data science, such as the use of predictive analytics to help guide the people making decisions. Here are a few examples (see the report ‘ Data Science for Local Government ‘ by the Oxford Internet Institute for further ones):
- In Leicester, emergency services are using predictive analytics to understand which types of people and parts of the city were putting the most demand on their services. These areas are then targeted with preventative measures (for instance, smoke-alarms) to ease the future strain on services.
- ‘Chatbots’, AI programs that intuitively interact online, are being used by councils across the UK to relieve pressure on telephone services. Calls that require human intervention can be screened, while other issues can be dealt with by Chatbots.
- Models built from geo-spatial data and representations like heat-maps have proved incredibly useful for visualising issues. For example, data science can be used to analyse crime patterns and help allocate policing resources, or inform decisions such as where future investment and services should be located.
What Next?
Data science brings huge opportunities. To realise its benefits there are things that local authorities must consider. Data science doesn’t magically clean data, replace the need for people, or remove potential biases and inequalities in existing datasets and processes.
Over the next seven articles we’ll show you how to set up data science projects that bring real benefits and avoid potential pitfalls, projects that are effective, intelligent and ethical — and make a real difference to citizens.
If you would like to find out more we’d be happy to discuss this topic with you, simply give us a call or drop us an email to set it up!
Download The Full Public Sector Playbook Here
Originally published at https://www.peakindicators.com on December 9, 2022.
