Why Data Mining?

Why Data Mining?

• Enormous data growth exits in commercial and scientific databases because of advances in data generation and collection technologies
• The new mantra is to gather whatever data possible, whenever and wherever as data has value
• Progressed from terabytes to petabytes of data

Why data mining? Commercial Viewpoint

• A lot of data being collected and warehoused, especially webdata
• Yahoo has petabytes of web data
• Facebook has billions of active users
• Amazon handles millions of visits per day
• Data from Purchases at department/grocery stores (loyalty cards), bank and credit card transactions is available
• Computers have become cheaper and more powerful enabling competitive pressures
• To create a competitive edge, businesses provide better, customized services to increase profit through Customer Relationship Management

Why data mining? Scientific Viewpoint

• Data can be collected and stored at enormous speeds from:
  • Remote sensors on a satellite
  • NASA EOSDIS archived >24 petabytes of earth science data during 2018
  • Telescopes scanning data
  • High-throughput biological data
  • Scientific simulations generating terabytes of data in a few
• Data mining helps scientists in automated analysis of massive datasets and hypothesis formation

Data mining at UKZN

• Moodle usage, student card swipes (RFID), and LAN logins
• Individuals can be tracked with smart/fitness watches and smartphones

McKinsey Global Institute report on Big Data

• Innovative competition and productivity is occurring because of big data
• $600 is the cost to buy a disk drive that can store all of the world's music
• 5 billion mobile phones were in use in 2010
• 30 billion pieces of content are shared on Facebook every month
• 40% projected growth in global data generated per year
• 5% growth in global IT spending
• 235 terabytes of data collected by US Library of Congress in April 2011
• 15 out of 17 sectors in the United States have more data stored per company than the US Library of Congress
• Potential Value of big Data:
  • $300 billion potential annual value to US health care.
  • €250 billion potential annual value to Europe's public sector administration
  • $600 billion potential annual consumer surplus from using personal location data globally
  • 60% potential increase in retailers' operating margins
• 140,000–190,000 more deep analytical talent positions and 1.5 million more data-savvy managers needed to take full advantage of big data in the United States

Solving Problems using Data

• Improving health care and reducing costs
• Predicting the impact of climate change
• Finding alternative/green energy sources
• Reducing hunger and poverty by increasing agriculture production

What is Data Mining?

• Data Mining,also known as knowledge discovery from data (KDD)
• Extraction of interesting (non-trivial, previously unknown and potentially useful) patterns, structure, or knowledge from huge amounts of data
• Exploration and analysis of data by automatic or semi-automatic means, to discover meaningful patterns/structure/knowledge
• The structure found may take many forms, including a set of rules, a graph or network, a tree, one or several equations, and more
• Data mining is explaining the past and predicting the future by means of data analysis

Data Mining Explained

• Data mining combines statistics, machine learning and database systems
• Traditional techniques are unsuitable with data that is Large-scale, High dimensional, Heterogeneous, Complex and Distributed
• Data mining is a key component of the emerging field of data science and data-driven discovery

Data Mining Steps

• Data cleaning (to remove noise and inconsistent data)
• Data integration (where multiple data sources may be combined)
• Data selection (where data relevant to the analysis task are retrieved from the database)
• Data transformation (where data are transformed or consolidated into forms appropriate for mining by performing summary or aggregation operations)
• Data mining (an essential process where intelligent methods are applied to extract data patterns)
• Pattern evaluation (to identify the truly interesting patterns representing knowledge based on some interesting measures)
• Knowledge presentation (where visualization and knowledge representation techniques are used to present the mined knowledge to the user)
• These steps form part of any data analytics cycle but are specific to data mining

What Data Mining is/is not

• Not data mining:
  • Look up phone number in phone directory
  • Query a Web search engine for information about "Amazon"
• Data mining:
  • Certain names are more prevalent in certain provinces (e.g., Dlamini, Zuma... in KZN)
  • Group together similar documents returned by search engine according to their context
• These are considered a data query, which is simply a request for data or information from a database table or combination of tables

Data Mining Tasks

• Two Categories of data mining tasks include:
  • Descriptive mining tasks characterize the general properties of the data
  • Predictive mining tasks perform inference on the data in order to make predictions
• Data mining functionalities are used to specify the kind of patterns to be found in data mining tasks
• There are data mining techniques used for each task to specify the kinds of patterns or structure to be found in data mining tasks

Data Mining Methods

• Prediction Methods can also involve estimation methods
• Target Outcome can be approximated.
• Data mining techniques examples:
  • Clustering : Grouping of records, observations, or cases into classes of that are similar
  • Association : Finding which attributes "go together."
  • Anomaly Detection : Detecting unusually large amounts for a given account number in comparison to regular charges incurred by the same account
  • Predictive Modeling : Model to predict an outcome for some target variable

Examples to Consider

• The main data mining techniques:
  • Data is clustered
  • Association rules
  • Predictive Modelling
  • Detecting Anomalies
• Common data mining techniques:
  • Predictive modeling is the process by which a model is created to predict an outcome for some target variable
  • If the outcome is categorical, it is called classification
  • If the outcome is numerical, it is called regression
  • Clustering refers to the grouping of records, observations, or cases into classes of similar objects
  • A cluster is a collection of records that are similar to one another, and dissimilar to records in other clusters Clustering differs from classification in that there is no target variable for clustering
  • Association techniques data mining are used for finding which attributes "go together"
  • The technique of association seeks to uncover rules for quantifying the relationship between two or more attributes in business known as with affinity analysis or market basket analysis.
  • Anomaly detection is also referred to as outlier detection or analysis.
  • These data objects are called outliers.

Data Mining Examples

• Determine whether or not each of the following activities is a data mining task:
  • Dividing the customers of a company according to their gender is a simple database query
  • Dividing the customers of a company according to their profitability is an accounting application of a new customer would be data mining
  • Computing the total sales of a company is simple accounting
  • Sorting a student database based on student identification numbers is simple database query
  • Predicting the outcomes of tossing a (fair) pair of dice is a probability calculation
  • Predicting the future stock price of a company using historical records is predictive modeling
  • Monitoring the heart rate of a patient for abnormalities is anomaly detection and even a classification problems
  • Monitoring seismic waves for earthquake activities is a classification problem

Machine Learning

• Machine learning (ML) is data analysis that automates analytical model building.
• In branch of artificial intelligence (AI), systems learn from data, identify patterns and make decisions with minimal human intervention without being explicitly programmed
• Machine learning focuses on developing computer programs that can access data and use it to learn to produce a program from data and output

Machine Learning - The Process

• Process of learning begins with observations or data, such as examples, direct experience, or instruction, in order to look for patterns in the data
• One of the more famous examples is a program that learned to recognize cats by being fed cat pictures
• Examples of where machine learning is currently used:
  • Image classification
  • Document classification
  • Hidden spam detection
  • Speed fraud detection
  • Digital advertising with AI to extract data
  • Self driving cars

Supervised Machine Learning Algorithms

• Apply past knowledge to new data using labelled examples to predict future events
• "Labelled data" is a designation for pieces of data that have been tagged with one or more labels identifying certain properties or characteristics, or classifications or contained objects
• Start with a training dataset and a learning algorithm produces a function to make predictions about the output values, so algorithms are trained on some correct examples
• Show an algorithm examples until it gets the idea and can recognize in other pictures

Reinforcement Learning Algorithms

• Learning method by environment interaction via actions and discovers errors or rewards
• Allows machines/software agents to automatically determine the ideal behavior in a context to maximize its performance with simple feedback to the agent to learn which action is best, the reinforcement signal
• Game example: by relating what buttons you press, to what happens on screen and your score, performance improves

Unsupervised Learning Algorithms

• No labelled data is involved at all and no feedback is delivered, so no reinforced learning is delivered
• Tasks algorithms identify patterns in data, to spot similarities that split that data into categories.
• Studies how systems infer a function to describe a hidden structure from unlabelled data, it explores the data and can draw inferences from datasets to describe hidden structures.
• Airbnb clustering houses available to rent by neighbourhood, and Google News grouping together stories on similar topics each day

Semi-Supervised Learning Algorithms

• Mixes supervised and unsupervised learning, where the algorithm is trained on both labelled and unlabelled data. Cat Photo example: The model can recognize a cat in any picture.

Why Use Machine Learning

• Some tasks cannot be defined well except by example
• Hidden among large piles of data are important relationships and correlations that are extracted by machine learning
• Machines can improve upon the original design in their operating environment
• The amount of knowledge available is too large to fully encode by humans so the machines learn so AI systems can conform to new knowledge
• Environments change over time, reducing the need for redesign
• Machine learning solves problems that cannot be solved by numerical means alone!