Estimates indicate that there'll be more than 21 billion IoT devices connected by 2025. That's a lot of data. In this lesson, we'll talk about this data. We'll begin by building a fundamental understanding of data types, IoT sensors, a few preliminary data processing techniques, and analytics. Data fundamentals. We can slice the pie in many ways. But broadly, there are two kinds of data. Quantitative and qualitative. Quantitative can be expressed as a number and measured. We can statistically manipulate and visualize quantitative data. Examples are cars driving speed, room temperature, etc. On the other hand, qualitative data cannot be expressed as a number nor be measured. Qualitative data can typically be sorted by category. Not by a number. Examples are our names, holiday destinations, etc. Now that we understand quantitative and qualitative, we'll understand two groups within each of these. Quantitative data can be discrete or continuous. Discrete data are absolute values and cannot be divided, like the number of employees in your organization. Continuous data can have decimal values and meaningfully divided into finer levels. Like there are millions of possible heights, such as 65.82342. You get the idea. Qualitative data can be nominal or ordinal. Nominal data are used to label variables that have no quantitative values. It has no order. If I ask you, are you married? The answer can be 1, yes or 2, no. Order did not matter. Ordinal data is nearly the same as nominal data, except it's ordering matters. If I ask, how are you feeling today? The answer can be 1, unhappy, 2, neutral, and 3, happy. It can also be used to measure other things, such as the degree of customer satisfaction. The most significant value of the IoT is that it creates information without the need for human observation. Sensors are electronic devices that can extract data from a physical condition or event. We'll now look at different types of sensors, their examples, and the data they produce. Position sensors, like a proximity sensor, measure the absolute or relative position of an object. Occupancy sensors detect the presence, whereas motion sensors detect people's moment or objects in an area. Examples include electric ice and radar. Velocity and acceleration sensors, as the name suggest, measures how fast something moves. Velocity sensors measure the speed of motion, whereas acceleration sensors measure the change in velocity. Examples include accelerometer and gyroscope. Force sensors, like force gauge, measure the magnitude of physical force. Pressure sensors, like barometers, measure the force applied by liquids or gases. Flow sensors, like anemometer, detect the rate of fluid flow. Acoustic sensors, like hydrophone, measure sound levels. Humidity sensors, like soil moisture sensors, detect humidity in the air or a mass. Light sensors, like infrared sensors, detect the presence of light. Radiation sensors, like neutron detectors, detect radiation in the environment. Temperature sensors, like thermometers, measure the amount of heat or cold present in the system. Chemical sensors, like smoke detectors, measure the concentration of chemicals in a system. Biosensors, like blood glucose biosensors, can detect various biological elements. We understand data and sensors now. The power of IoT comes from what we do with this data. These insights and actions can be generated in three areas. On edge, in the fog, or the cloud. Determining the edginess depends on data urgency, value, network bandwidth, etc. Let's understand the qualities of these three areas, their benefits, and examples. In an edge, the actions taken near the thing, which could be a device or the sensor, you can act in real time to imminent threats. An example is avoiding accidents in a self-driving car. In the fog, the actions taken between the edge and the cloud. Fog is transactional in nature and you can react to triggering events based on it. An example is when your bank sends you a fraudulent transaction alert. In the cloud, the action is taken in the core enterprise system. Data is informational in nature. An example is optimizing maintenance schedules based on historical data. Once data collection strategy and standards have been defined, we can begin to process it through different analytics levels. There are four levels of analytics based on its business value and complexity deriving it. These analytics maturity levels are descriptive, diagnostic, predictive, and prescriptive. Descriptive analytics are the building blocks of an IoT analysis program. It describes what happened with the devices in the connected flip, what is the average uptime versus downtime of a machine for the past month, what is the device's maximum runtime each day. The second level, diagnostic analytics, digs deeper into data provided by descriptive analytics. It can detect anomalies in the device behavior and help determine its root cause, what contributed to device failure, is there a correlation between the operating environment and device failure, etc. The third level, predictive analytics, uses linkages between past device behavior and events to project future device behaviors. The benefits of this prediction are endless. It can help schedule proactive machine maintenance, limit downtime, operate effectively, and ultimately saves hundreds and thousands of dollars. The pinnacle of analytics, prescriptive analytics leaps from predicting future device behavior to automatically forcing a device behavior based on business rules. A self-driving car is a great example. It's analytics models can assess different road driving conditions, prescribe a range of suggested actions, and even execute it.
