Fundamentals of Machine Learning
Prepared by
Prince Thomas
Chapter 1: Introduction to Machine Learning
Artificial Intelligence and Machine Learning
- What is Artificial Intelligence?
- Relationship between AI, Machine Learning, and Data Science
What is Machine Learning?
- Definition and Key Concepts (Features, Labels, Models, and Predictions)
- Machine Learning Pipeline (Data Collection, Preprocessing, Model
Training)
Machine Applications
- Applications in Healthcare, Agriculture, Finance, and Robotics
Types of Machine Learning
- Supervised Learning
- Unsupervised Learning
- Reinforcement Learning
Artificial Intelligence and Machine Learning
What is Artificial Intelligence (AI)?
• AI aims to develop machines that can mimic human cognitive
functions like decision-making, problem-solving, and language
understanding.
• AI includes various fields such as Natural Language Processing
(NLP), Computer Vision, Robotics, Machine Learning and Deep
Machine learning enables a machine to automatically learn from data,
improve performance from experiences, and predict things without being
Relationship between AI, ML, and Data Science
• AI is the broader field covering any machine intelligence.
• ML is a subset of AI where systems improve through experience
with data.
• Data Science involves collecting, cleaning, analyzing, and
What is Machine Learning?
Definition:
• Machine Learning enables computers to automatically learn and
adapt from experience without being explicitly programmed for
every task.
Classification is predicting the category or class of a new data based on its features.
Answers gives exact number then this type of problem statements are called
Regression.
Clustering is Grouping similar data points together based on their inherent
characteristics(here based on colour and shape) or patterns.
Key Concepts of Machine Learning:
• Features: The input variables or characteristics used by the
model (e.g., patient age and blood pressure).
• Labels: The output or target value the model predicts (e.g.,
‘Disease Present’ or ‘Not Present’).
• Model: A trained mathematical system (e.g., linear regression
model or neural network) that processes features to generate
predictions.
• Predictions: The final output produced by the model based on
Image classification:
Each image is labeled with a corresponding object or scene (e.g., "cat," "dog,"
"mountain").
Sentiment analysis:
How does Machine Learning works
Cont.,
Machine learning works by developing algorithms and models that can learn from
data and make predictions or decisions based on that data. The basic steps involved
in machine learning are as follows:
1. Data collection: The first step in machine learning is to collect a large, diverse,
and representative dataset of examples, along with some desired output or outcome
for each example.
2. Data pre-processing: The next step is to preprocess the data to prepare it for
use in the machine learning algorithm. This may involve tasks such as cleaning the
data, removing outliers, and normalizing the data.
3. Feature engineering: Feature engineering involves selecting or extracting
relevant features or attributes from the data that can be used to make predictions or
decisions. This may involve tasks such as selecting the most important variables,
Cont.,
4. Model selection: The next step is to select an appropriate machine
learning model or algorithm that is well-suited to the problem at hand. This
may involve tasks such as selecting a classification algorithm, regression
algorithm, or clustering algorithm.
5. Model training: Once a model has been selected, the next step is to train
the model on the dataset. The model is adjusted or optimized to minimize the
error or loss between the predicted output and the desired output.
6. Model evaluation: After training, the model is evaluated on a separate
validation or test dataset to assess its performance and generalization ability.
The model may be refined or adjusted based on the evaluation results.
7. Model deployment: Once the model has been trained and evaluated, it
Dataset
Original Dataset
This is the complete dataset collected for the ML project.
Data Splitting
Training Set: Used to train the machine learning model
by fitting the algorithm to the data. Typically 60-80% of
the training set.
Validation Set: Used to tune hyperparameters (e.g., learning
rate) and prevent overfitting. Usually 10-20% of the original dataset.
Test Set: Used to evaluate the final performance of the trained model. The
model is tested on this unseen data to assess its generalization ability.
Typically 10-20% of the original dataset.
Applications of Machine learning
Types of Machine Learning
1. Supervised Learning
• In supervised learning, models learn from labeled datasets, where each data
point contains both features (input) and corresponding labels (correct
output).
• The objective is to predict the label for new, unseen data points using the
patterns learned from the training data.
Example: Predicting Student Grades:
• Features: Study hours, attendance
• Label: Previous exam result (grade)
• The model learns how study habits influence exam performance and
Common Algorithms:
• Linear Regression: Predicts continuous values (e.g., house prices).
• Decision Trees: Splits data based on features to classify or predict
outcomes.
• Neural Networks: Complex models inspired by the brain, useful for both
regression and classification tasks (e.g., image recognition).
Applications:
• Healthcare: Disease prediction (e.g., predicting diabetes).
• Finance: Credit risk assessment.
2. Unsupervised Learning
In unsupervised learning, models work with unlabeled data, meaning the
dataset contains only input features without predefined labels.
The model’s goal is to discover hidden patterns, structures, or relationships
within the data.
Example: Customer Segmentation:
Using data on purchasing behavior (e.g., frequency of shopping, spending
amount), the model groups customers into clusters to find patterns like
“frequent shoppers” or “seasonal buyers.”
Common Algorithms:
K-Means Clustering: Groups data points into clusters based on feature
similarity.
Principal Component Analysis (PCA): Reduces dimensionality of data
while retaining essential patterns.
Applications:
Retail: Customer segmentation for personalized marketing.
Finance: Detecting unusual patterns in transactions (fraud detection).
Healthcare: Grouping patients based on symptoms for better diagnosis.
3. Reinforcement Learning
Reinforcement learning (RL) is a type of machine learning where an agent
learns through interactions with the environment and receives rewards or
penalties for its actions.
The goal is to develop an optimal strategy (policy) that maximizes cumulative
rewards over time.
Example: Training a Robot to Walk:
The robot takes steps, receives rewards for maintaining balance, and
penalties for falling.
Over time, the robot learns which actions maximize its stability and
Common Algorithms:
Q-Learning: An RL algorithm that uses a Q-table to store and update the
value of actions in each state.
Deep Q-Network (DQN): Combines deep learning with Q-learning to handle
large, complex environments.
Applications:
Video Games: AI agents playing games like chess or Go.
Self-Driving Cars: Learning to navigate traffic safely.
Robotic Control: Training robots for industrial tasks (e.g., assembling parts).
