Chapter 1
Introduction to Machine Learning
Need For Machine Learning
• Business organizations use huge amount of data for their daily
activities
• Earlier this data was not utilized at full potential
– Two reasons
• Data being scattered at different archives and not able to integrate
• Lack of software tools to unearth useful information
• Popularity of machine learning
– High volume of available data to manage :FB, X (twitter), YouTube.
Gets doubled/year
– Cost of storage has reduced: cheap h/w, easy to capture, process,
store, distribute and transmit
– Availability of complex algorithms
The knowledge pyramid
• Data : All facts are data. Text, audio,
image, video etc
• Information: processed data. Include
patterns, associations/relationships
among data.
– Eg. Sales data can be analyzed to extract
information like which is the fast moving/selling
product
• Knowledge : condensed information
– Historical patterns and future trends of the above
sales example is called knowledge
• Intelligence: applied knowledge for
action /an actionable form of knowledge
• Wisdom: that represents maturity of
mind. So far, exhibited by humans only !!!
Machine learning explained
• Machine learning is important sub branch of AI
• Arthur Samuel. Pioneers of AI, “ machine learning is the field of study that gives
the computers the ability to learn without being explicitly programmed ”
– System should learn by itself without explicit programming
• Conventional programming uses predefined rules or logic to solve a problem, while
machine learning uses data to construct a model or logic for the problem
• Conventional programming relies on programming logic, while machine learning
depends on real-life data-driven approaches that involve learning and modeling.
• Conventional programming is a manual and fixed process, where the programmer
designs the solution explicitly, while machine learning is an automatic and flexible
process, where the algorithm learns from data to generate a solution.
• As humans take decisions based on an experience, computers make
models based on extracted patterns in the input data and then use these
data data-filled models for prediction and to take decisions.
What is model?
• A model can any one of the following
– Mathematical equation
– Relational diagrams like trees/graphs
– Logical if/else rules
– Groupings called clusters
• A model can be a formula, procedure or representation that can generate
data decisions
Pattern Model
Local and only applicable to Global and fits the entire
certain attributes dataset