IoT Data Analysis

An Internet of Things (IoT) architecture has devices, network connecting these devices, platforms that receives the data and take actions, and Data Analysis (Analytics), which give intelligence IoT networks.

In insidebiggata we can see a very interesting vision of Analytics in IoT which are summarized in this article:


IoT has a lot of data, larger and faster continually receiving them, but the IoT no means to receive and store the data, it must learn to analyze them (hence the Analytics) to give a competitive advantage and business systems.


With its massive collection of data, you can make a better decision-making, from now with IoT data sources quickly generate, therefore handling such fast data is the challenge. With analysis tools such as streaming Streaming Spark, will be able to analyze and take action quickly, it will end with the heavy task of receiving, storing, managing and analyzing data so heavy.


When it comes to real-time, you should talk time business, when the data is  received and that will be necessary to take some action or other, for example, a customer who is in a shopping center data, the actions will be different depending on the day and time will not be the same actions a day of sales on Christmas day.


4 types of analysis on the IoT defined and any solution must attack the four approaches:

  • Descriptive analysis: Based on the traditional analysis, the data are collected, are displayed on a panel, reports are drawn and displayed in a tool to use. With this we see what is happening and can deduce behaviors and patterns. It has human vision of the information we have.
  • Predictive analysis: With the data collected is learned and predictive statistical models based on experience tell us will happen are removed.
  • Analytical streaming : Here patterns, events and data on which to act is seeking, but you can not get to this model without the previous two (descriptive and predictive) without human vision of information because they do not know what you want or not predictive that can not automate the process if you do not know what will happen.
  • Prescriptive Analytics : We already have the pattern or event with this information, we must make an analysis based on restrictions of the company, some other predictive model and company policies to determine what actions to take.

Another point to consider is the philosophy of Internet of Things (IoT) architectures, these increasingly have and will have more devices with tremendous growth of users, devices, infrastructure and data, because this is increasingly being thinking more on analysis on the edge (edge) against the traditional analysis in a centralized database with large amounts of data, there will be an immediate Analytics to take action directly from the edge, we must decide what data you can manage there and which will have to take them to the centralized system (without saturating it ) and Analytics applications'd better run on the Edge or centralized, taking into account latency networks, capacities, delays, security ... .. 



Social Network :