One of the biggest trends affecting the analytics sector at the moment is the emergence of the Internet of Things (IoT) as a key source of data.
Over the next few years, the number of IoT devices is set to explode. By the end of 2020, Juniper Research forecasts there will be some 38 billion such items in use, a threefold increase since 2015.
But while this will present huge new opportunities for businesses to apply big data analytics to the information generated in order to gain valuable insight, it does pose a range of risk as well.
Although questions such as privacy are well documented, one issue that is frequently overlooked is the quality of the data itself. Businesses may assume that because the incoming data will be taken directly from sensors, there will be little that can go wrong with it, but in fact, this is not necessarily the case.
It was noted by Mary Shacklett, president of Transworld Data, that one issue that may frequently affect the quality of IoT data lies in fundamental flaws in the way the embedded software used in the devices is developed.
She explained in an article for Tech Republic that historically, developers of this software – which runs machines, produces machine automation, and enables machines to talk to one another – did not always employ the same methods as they would for more traditional apps.
"This meant that detailed quality assurance (QA) testing on the programs, or ensuring that program upgrades were administered to all machines or products out in the field, didn't always occur," Ms Shacklett stated.
The result of this could be significant for big data operations. If an undetected flaw in an IoT device's embedded software results in inaccurate data being generated, this could lead to an erroneous analytics conclusion that has a major impact on the business.
Although this is changing as more manufacturers mandate strict compliance and QA testing from their embedded software developers – with sectors such as automotive, aerospace, and medical equipment leading the way due to their high quality standards, for now, this remains a risk that must be considered when using IoT data.
To counter this, Ms Shacklett highlighted two key steps to ensure the quality of this information. Firstly, she noted that users must monitor their generated data closely, and immediately investigate any unusual readings, which also need to be reported to the appropriate teams.
For instance, "if the team charged with end responsibility for machines/devices sees anything unusual with the data, immediate action should be taken on the floor, and they must report back to the analytics team that a potential problem could affect data".
Organisations also need to ensure that vendors are kept in the loop, on both the analytics and machine side. It was noted that as hardware and software is never perfect, there may be some instances where data might be skewed by a known issue that a machine manufacturer or IoT provider is experiencing.