Medway Youth Trust, a charity for young people in Kent, has invested £100,000 on IBM SPSS Predictive Analytics software, which helps it identify young people at risk of unemployment after leaving school, so the charity can advise them on employment, higher education or training much earlier on.
Established in 2008, Medway Youth Trust helped approximately 24,500 13-21 year olds in 2011, about 1,000 of whom came into the youth centre of their own accord. Youth centre advisors offered one-to-one guidance on health, housing, employment and financial matters.
Gary Seaman, data quality manager at Medway Youth Trust, said the analytics was initially used for updating the employment status of the charity's young people.
"We report certain criteria to the Department of Education on a monthly basis, which feeds into the national employment figures. If young people are not in training, employment or education, they are flagged as unemployed.
"However, looking through a list of some 3,500 young people and ensuring their [data was typed up and updated correctly] was time-consuming.
"We looked into software that could read text against a structured data field so we could correct data anomalies more quickly," he said.
The trust then looked into whether the same data could be used to predict what young people would go on to do after leaving school.
"We wanted to predict which young person was more likely to become a NEET [not in employment, education or training] after school and what factors determine this, that way we could make an early intervention," he said.
Graham Clewes, CEO of Medway Youth Trust, said the IBM SPS Predictive Analytics software was chosen instead of SAS and SAP offerings because of "the user-friendly nature of the software but, most importantly, the ability to extract data from multiple data sets that are not compatible with each other".
This was important because the charity works with 30 partners, which meant 30 different databases that "didn't talk to each other at all".
Seaman explained how the software works. "Our advisors speak to young people daily and input information onto the system in free text form. The IBM software prepares the data and filters out irrelevant data.
"We extract the textual data and update the data dictionary. The prepared structured data and the unstructured data are joined together and we add data from other data sets. From there the modelling process uses complex algorithms that make up the back end of the modelling stage," he said.
Seaman says a big advantage of using the IBM modeller part of the SPSS software package is that it is automated.
"If I want to work out the risk of NEET, the process is automated. It analyses all the data sets we have and attaches the original source data. Finally, we run an Excel CSV file with the client name, number and propensity score, which is imported back into our secure Career Vision CRM database," he said.
In August 2010, IBM predictive analytics software was installed on the charity's systems. It took a few hours to get the software up and running, and Seaman says he and the IBM consultant started analysing the data straight away.
"On day one, the consultant and I talked through the key things we needed from the data. The majority of the time taken was in building up the data dictionary as we had to categorise words from the 175 million we had. This process took two weeks and I am 90 per cent confident the data is as rich as possible," he said.
After the installation, Seaman said Medway Youth Trust bought 27 IBM consultancy days to help it build the initial NEET propensity model.
Seaman also had a three-day training course with IBM on the modelling, which included how to use the software, test the NEET prediction model, and tweak and improve it, and a further two days of training in text analytics.
Clewes said the news of the software wasn't released until the charity knew it had benefited from its use.
"Although the project took 57 weeks and was completed in October 2011, the process of analytics does not take that long. We received evidence that the analytics software was working in week six.
Clewes said that 52 per cent of the 723 young people identified by the analytics software were in employment, education or training after leaving school. He said the result was positive, but the fact that 48 per cent of them were still considered NEET showed there is more work to be done.
The trust may now look to use predictive analytics to help local communities in other ways, such as identifying young people more likely to use drugs or have early pregnancies.
Clewes says the charity presented its case study with IBM to 34 local communities, many of which are interested in deploying the software to help their local areas.
The charity spent four per cent of its annual expenditure – £100,000 – on the project, which included the software and consultancy expenses.
By eliminating high entry costs for big data analysis, you can convert more raw data into valuable business insight.
A discussion of the "risk perception gap", its implications and how it can be closed