IoT Analytics

Understanding IoT Analytics

Collecting sensor data and transforming it into meaningful information allows you to understand how your “thing” is used in the real world. It forms the foundation of your product and service improvement, customer insights and identifying true customer value.

If you are new to the topic of analytics it can be confusing, so I will attempt to give a simple framework to put it into context.

future-present-past-choice-29624927Once you have collected the data of your sensors it’s time to do something useful with the numbers. In manufacturing it may be reducing maintenance or improving performance. I retail it could be looking at customer behavior or targeted advertising, in health it may be fitness improvement of monitoring of a condition and in cities it could be for efficient running of the transport system or energy saving.

I find it helpful to group the analytics into three time frames:

PAST

Descriptive analytics – You need to build a history of data to enable you to look into the past to understanding what has happened – Often information is visualized to allow humans to identify historic patterns and trends.

Diagnostic analytics – Problem solving or process optimization.

PRESENT

Real-time analysis – Looks at live data and often involves complex calculations.

Rules engines – Normally looks at data coming in against a set of rules e.g. if temperature is above 21 deg turn down the thermostat.

FUTURE

Predictive analytics – Matching historic patterns/models against current data that caused a particular result and assuming this will happen again.

Prescriptive analytics – This uses predictive analytics and then makes changes to the system it is connected to to alter the outcome (closed loop).

Insights – Hopefully from all the data collected you will be able to gain real insights into how customers  use and react to your products and services. Combine this with additional data sets on the Internet for the same time periods and you may be able to see patterns and behaviors that were impossible to identify in any other way, leading to the development of new business models, products and services.

Sensors are important

Temp_humi_pro

Analytics by design – Identifying in advance the full range of data you may want to collect is VERY important, even if you decide not to use all the data in the initial roll-out of your product/service.

Why? – Sensors are relatively cheap to buy and the incremental cost of adding an extra sensor during the initial design of the product is low cost.

However, the real costs are in the deployment. If you forget to include a crucial sensor in your initial product / service design the cost to retrofit the device and associated software upgrades may be very expensive, or even impossible if your sensor is embedded or located in a hazardous location.

It’s well worth taking extra time in the ideation phase to determine all the information requirements you may need to help you create value in the future (Remember the real value is in the data).