ML for Structured Data

— Week 1—
Fall 2025

Dr. M. Shahid Iqbal

1
 Recommended Background
 The formal prerequisite for this course is an intro to
programming, database systems, and some familiarity with
Probability and Statistics.
 If you don’t have background in these areas, you may still
sign up, but be aware that you will probably need to learn
some of these concepts as the class goes on (we will be
providing pointers to references)
 Course Detail
Text Books:
 Machine Learning, Tom, M., 1st Edition (1997)
McGraw Hill
 Hands-On Machine Learning with Scikit-Learn and
TensorFlow Concepts, Tools, and Techniques to
Build Intelligent Systems Beijing Boston, by Aurélien
Géron (Atleast first edition)
 Introduction to Machine Learning, Ethem Alpaydin,
2nd Edition (2010) MIT Press
 Grading Policy
1. Grading details:
 Four Assignments (at least) (15 Marks)
 Four Quizzes (at least) (10 Marks)
 May be a course project but not confirmed
 Mid Term (25 marks)
 Final Term (50 Marks)

2. Try your best to follow the deadlines and avoid copying.

3. Be punctual in the class.
4. Questions are always welcome.
 Introduction
 What Is Machine Learning?
 What is the difference between Data Mining and Data
Science?
 What is Datafication?
 Structure vs Unstructured data?
 Types of ML
 Difference between classification and regression?
 A real-life example
 Why Use Machine Learning?
5
 What is Machine Learning
 The field of machine learning is concerned with the
question of how to construct computer programs that
automatically improve with experience (T. Mitchell).
OR
 Principles, methods, and algorithms for learning and
prediction on the basis of past experience.
OR
 In the broadest sense, any method that incorporates
information from training samples in the design of a
classifier employs learning.

6
 What is Data Mining?
 Many Definitions
 Non-trivial extraction of implicit, previously unknown
and potentially useful information from data
 Exploration & analysis, by automatic or semi-automatic
means, of large quantities of data in order to discover
meaningful patterns
 Extraction of interesting (non-trivial, implicit, previously
unknown and potentially useful) patterns or knowledge
from huge amount of data (Knowledge discovery from
data

7
 What is Data Science
 Data science is lying at the intersection of computer
science, statistics, and substantive application domains.
 It is the study of “Where information comes from, what it
represents and how it can be turned into a valuable
resource in the creation of business and IT strategies”
 Data science is a "concept to unify/merge statistics,
data analysis, machine learning and their related
methods" in order to "understand and analyze actual
phenomena" with data”.

8
 Difference b/w Data Mining and Data
Science
Data Mining Data Science

Data mining is a process of extracting Data science refers to the process of

useful information, patterns, and trends obtaining valuable insights from structured
from huge databases. and unstructured data by using various tools
and methods.

Data mining is a technique. Data science is a field.

Primarily used for business purposes. Primarily used for scientific purposes.

It is involved with the process. It emphasizes the science of the data.

Data mining aims to make data more The objective of data science is to create a
important and usable; it means extracting dominant data product.
only useful information.

Data mining is a technique that is a part of It is related to the field of study like
KDD (Knowledge discovery in database Mechanical engineering, Cloud architecture,
process). etc.

It primarily deals with structured data. It deals with any kind of data like structured,
semi-structured, and unstructured.

9
 Datafication
 When we browse the web, we are unintentionally, being
datafied through cookies that we might or might not be
aware of.

 When we walk around in a store, or even on the street,

we are being datafied in a completely unintentional way,
via sensors, cameras, or Google glasses.

10
 Properties of Data

 Structured vs. Unstructured Data.

 Quantitative vs. Categorical Data

11
 What is Structured Data?
 Data organized in rows and columns
 Stored in relational databases (RDBMS)
 Easy to search, query, and analyze with SQL
 Examples: transaction records, customer details, sensor
logs
 Visual Suggestion: Screenshot of a table (columns: ID,
Name, Age, Salary) to depict structured format.

12
 What is Unstructured Data?
 Data not organized in predefined format
 Lacks rows/columns → difficult to analyze directly
 Examples:
 Text: emails, social media posts, articles
 Multimedia: images, audio, video
 Logs: web clickstreams, chat conversations
 Requires NLP, computer vision, or advanced ML for
analysis
 Visual Suggestion: Word cloud + image + video icons.

13
 Different Types of Structured Data
 Numeric Data: integers, decimals, financial data
 Categorical Data: gender, country, product type
 Date/Time Data: timestamps, event logs
 Ordinal Data: rankings (e.g., survey responses: poor →
excellent)
 Boolean Data: true/false, yes/no values
 Visual Suggestion: Table with columns → Type |
Example

14
 Quantitative vs. Categorical Data
 Quantitative data: Consists of numerical values, like
height and weight.
 Such data can be incorporated directly into algebraic
formulas and mathematical models, or can be displayed in
conventional graphs and charts.

 Categorical data: Consists of categories describing the

properties of the objects under investigation, like gender,
hair color, and occupation.
 Categorical data can usually be coded numerically. E.g.
gender might be represented as male = 0 and female = 1.

15
 Types of Machine Learning
Supervised learning, where we get a set of training inputs
and outputs. The correct output for the training samples is
available

Unsupervised learning, where we are interested in

capturing inherent organization in the data. No specific
output values are supplied with the learning patterns

Reinforcement learning, where there are no exact outputs

supplied, but there is a reward (reinforcement) for
desirable behaviour

16
 Classification and Regression
 Classification: To assign a label to an item from a
discrete set of possibilities.
 Such problems as predicting the winner of a particular
sporting contest (team A or team B?) or deciding the
genre of a given movie (comedy, drama, or animation?)
are classification problems.

17
 Classification and Regression
 Regression: To forecast a given numerical quantity.
 Predicting a person's weight or how much snow we will
get this year is a regression problem.
 Where we forecast the future value of a numerical
function in terms of previous values and other relevant
features.

18
 Problem Analysis

Given a real-life problem

19
 Problem Analysis
 Set up a camera and take some training samples to
extract features

Length

Lightness

Width

Number and shape of fins

Position of the mouth, etc…
 This is the set of all suggested features to explore for use
in our classifier!
Purpose:
To classify the future samples based on the data of
extracted features from the training samples
11/06/25 20
 An Example
A classification problem: the grades of students taking this
course

Key Steps:
1. Data (what past experience can we rely on?)
2. Assumptions (what can we assume about the students or
the course?)
3. Representation (how do we “summarize” a student?)
4. Estimation (how do we construct mapping from students
to grades?)
5. Evaluation (how well are we predicting?)
6. Model Selection (perhaps we can do even better?)

21
 An Example
1. Data: The data we have available may be:

- names and grades of students in past years ML courses

- academic record of past and current students

Student ML Course X Course Y

Peter A B A Training
David B A A data

Jack ? C A Current/Future
Kate ? A A data

22
 An Example

2. Assumptions:

There are many assumptions we can make to facilitate

predictions

1. The course has remained same roughly over the years

2. Each student performs independently from others

23
 An Example

3. Representation:

Academic records are rather diverse so we might limit the

summaries to select few courses
For example, we can summarize the ith student (say Bilal) with a
vector
Xi = [A C B]
where the grades may correspond to numerical values

24
 An Example

3. Representation:

The available data in this representation is:

Training data Data for prediction

Student ML grade Student ML grade (Class)
X1t B X1p ?
X2t A X2p ?
… …

25
 An Example

4. Estimation

Given the training data

Student ML grade
X1t B
X2t A
…

We need to find a mapping from “input vectors” x to “labels” y

encoding the grades for the ML course.

26
 An Example

Possible solution (nearest neighbor classifier):

1. For any student x, find the “closest” student xi in the training

set
2. Predict yi, the grade of the closest student

27
 An Example

5. Evaluation

How can we tell, how good our predictions are?

- we can wait untill the end of this course...
- we can try to assess the accuracy based on the data
we already have (training data)

Possible solution:
- divide the training set further into training and test
sets
- evaluate the classifier constructed on the basis of
the training set on the test set

28
 An Example

6. Model Selection

• We can refine
- The estimation algorithm (e.g., using a classifier
other than the nearest neighbor classifier)
- The representation (e.g., base the summaries on a
different set of courses)
- The assumptions (e.g., perhaps students work in
groups) etc.
• We have to rely on the method of evaluating the accuracy

of our predictions to select among the possible refinements

29
 An Example

Data can be
- Symbolic or Categorical (e.g. High Temperature)
- Numerical (e.g. 45 0C)

We will deal with both Symbolic and Numerical data

30
 An Example

From the available data we can

- Model the system which has generated the data
- Find interesting patterns in the data

There are multiple modelling approaches. i.e. Rule based

modelling of the system from which the data was generated

The search for interesting patterns is considered to be the

domain of Data Mining

31
 Why Use Machine Learning

First, there are problems for which there exist no

human experts.

Example: in modern automated manufacturing facilities,

there is a need to predict machine failures before they
occur by analyzing sensor readings. Because the
machines are new, there are no human experts who can
be interviewed by a programmer to provide the knowledge
necessary to build a computer system. A machine learning
system can study recorded data and subsequent machine
failures and learn prediction rules.

32
 Why Use Machine Learning

Second, there are problems where human experts

exist, but they are unable to explain their expertise.

This is the case in many perceptual tasks, such as speech

recognition, hand-writing recognition, and natural language
understanding. Virtually all humans exhibit expert-level
abilities on these tasks, but none of them can describe the
detailed steps that they follow as they perform them.
Fortunately, humans can provide machines with examples
of the inputs and correct outputs for these tasks, so
machine learning algorithms can learn to map the inputs to
the outputs.

33
Why Use Machine Learning

34
 Why Use Machine Learning

Third, there are problems where phenomena are

changing rapidly.

Example: people would like to predict the future behavior

of the stock market, of consumer purchases, or of
exchange rates. The rules and parameters governing
these behaviors change frequently, so that the computer
program for prediction would need to be rewritten
frequently.

35
 Why Use Machine Learning

Fourth, there are applications that

need to be customized for each
computer user separately.

Example: a program to filter

unwanted electronic mail messages.
Different users will need different
filters.

36
Any Question

37