Big Data
Over the past year there have been growing reports of the phenomenon called big data. Through clever statistical analysis of large data sets some say you can start to predict future events.
Earlier this year the BBC's Horizon programme looked at an approach being trialled in the US to predict the locations and times of crimes. If we could predict when anti-social behaviour was going to happen or when a repair was needed this must surely be a good thing. So is big data something the housing sector should start to explore?
Maybe, but there are other things that can produce more value. When we think about big data our minds turn to large data bases of housing stock. However, even those of the largest social landlords — containing information on around 100,000 properties — would be at the very bottom end of the big data world. While this data is valuable, in the short term we can also use small data and the humble Excel spreadsheet in interesting ways.
Housing associations generate large amounts of data every day, but only a small fraction is analysed to try to improve the business.
There are several reasons why we often fail to transform data into useful information. A key one is the diverse range of skills that need to be combined to carry out productive analysis. You need a good understanding of business needs, an appreciation of how to use data, knowledge of statistical analysis and a good dose of commercial common sense.
An area that analysing small data can be useful in is in response repairs, one of the largest costs associations have to bear. Great efforts are made to get the best value for money but the focus tends to be on the supply side, with every effort taken to get the best cost from contractors or in-house repair teams.
We should also focus on the demand side of the equation, and get an understanding of how it is driving costs.
How to understand and reduce costs using small data
Look at what proportion of your repair budget in a year is spent on the 20% of stock that has the highest per unit cost in the year. This can be done quite easily on an Excel spreadsheet and the results might surprise you. Having looked at two organizations, the rate has ranged between 48% to 80%.
In other words, in the extreme case 80% of the budget was being spent on 20% of the stock. In that example the remaining 20% of the budget was being spent on 60% of the stock and 20% of the stock was getting nothing at all.
From this you might conclude that you should sell the stock that is eating up so much of your repairs budget. But if you analyse two years' of information you may find that it is not always the same properties in the 20%. When I did this, I found that only 5% of the stock was in the top 20% in two consecutive years.
If you discover figures such as these, you might want to move to the second phase: making the information visible. List which homes are costing the most to repair. This should be a standard report for property management staff so that both the issue and the cost are visible and then make it someone's job to find out why the costs are structured in the way they are.
You can do this through statistical analysis to look at whether factors such as age, location, property type, tenancy type, and so forth, shows any significant correlation with the cost of repairs, or make it someone's job to go to the homes that are costing the most money, find out why and stop it. If resource precludes both, do the latter.
If big data makes us all more aware of the importance and value that can come from analysis of data it will have done a good thing, but we should look at what we have already and think a bit harder.
No comments:
Post a Comment