The value of IoT can be found within the analysis of data gathered from the system under observation, where insights gained can have direct impact on business and operational transformation. Through analysis data correlation, patterns, trends, and other insight are discovered. Insight leads to better communication between stakeholders, or actionable insights, which can be used to raise alerts or send commands, back to IoT devices. With a focus on the topic of Exploratory Data Analysis, the course provides an in-depth look at mathematical foundations of basic statistical measures, and how they can be used in conjunction with advanced charting libraries to make use of the world’s best pattern recognition system – the human brain.
Learn how to work with the data, and depict it in ways that support visual inspections, and derive to inferences about the data. Identify interesting characteristics, patterns, trends, deviations or inconsistencies, and potential outliers. The goal is that you are able to implement end-to-end analytic workflows at scale, from data acquisition to actionable insights.
Through a series of lectures and exercises students get the needed skills to perform such analysis on any data, although we clearly focus on IoT Sensor Event Data.
After completing this course, you will be able to:
• Describe how basic statistical measures, are used to reveal patterns within the data
• Recognize data characteristics, patterns, trends, deviations or inconsistencies, and potential outliers.
• Identify useful techniques for working with big data such as dimension reduction and feature selection methods
• Use advanced tools and charting libraries to:
o Automatically store data from IoT device(s)
o improve efficiency of analysis of big-data with partitioning and parallel analysis
o Visualize the data in an number of 2D and 3D formats (Box Plot, Run Chart, Scatter Plot, Pareto Chart, and Multidimensional Scaling)
For successful completion of the course, the following prerequisites are recommended:
• Basic programming skills in any programming language (python preferred)
• A good grasp of basic algebra and algebraic equations
• (optional) “A developer’s guide to the Internet of Things (IoT)” – a Coursera course
• Basic SQL is a plus
In order to complete this course, the following technologies will be used:
(These technologies are introduced in the course as necessary so no previous knowledge is required.)
• IBM Watson IoT Platform (MQTT Message Broker as a Service, Device Management and Operational Rule Engine)
• IBM Bluemix (Open Standard Platform Cloud)
• Cloudant NoSQL (Apache CouchDB)
• Languages: R, Scala and Python (focus on Python)
Introduction to exploratory analysis
Analysis of data starts with a hypothesis and through exploration, those hypothesis are tested. Exploratory analysis in IoT considers large amounts of data, past or current, from multiple sources and summarizes its main characteristics. Data is strategically inspected, cleaned, and models are created with the purpose of gaining insight, predicting future data, and supporting decision making. This learning module introduces methods for turning raw IoT data into insight.
Graded: Challenges, terminology, methods and technology
Graded: Week 1 Programming Assignment 1
Graded: Week 1 Programming Assignment 2
Tools that support IoT solutions
Data analysis for IoT indicates that you have to build a solution for performing scalable analytics, on a large amount of data that arrives in great volumes and velocity. Such a solution needs to be supported by a number of tools. This module introduces common and popular tools, and highlights how they help data analyst produce viable end-to-end solutions.
Graded: Data storage solutions, and ApacheSpark
Graded: Programming language options and functional programming
Graded: ApacheSparkSQL, Cloudant, and the End to End Scenario
Graded: Week 2 Programming Assignment
Mathematical Foundations on Exploratory Data Analysis
This learning module explores mathematical foundations supporting Exploratory Data Analysis (EDA) techniques.
Graded: Averages and standard deviation
Graded: Skewness and kurtosis
Graded: Covariance, correlation and multidimensional Vector Spaces
Graded: Programming Assignment 3
This learning module details a variety of methods for plotting IoT time series sensor data using different methods in order to gain insights of hidden patterns in your data.
Graded: Visualization and dimension reduction
Graded: Programming Assignment Week 4
ENROLL IN COURSE