Coding Club Coding Week 25

IIT Guwahati

Machine
Learning
Resources Released
Machine Learning Resources

What is Machine Learning?

Machine Learning (ML) is a dynamic and rapidly growing subfield of Artificial Intelligence (AI)

that focuses on enabling computers to learn from data and make informed decisions without

being explicitly programmed. It allows systems to improve their performance over time by

recognizing patterns and drawing insights from vast amounts of information.

Today, ML is at the core of many cutting-edge technologies — from image and speech

recognition to language translation, recommendation systems, financial modelling, and

autonomous vehicles. Its impact is widespread across industries including healthcare, finance,

e-commerce, entertainment, and robotics.



As a foundational pillar of modern AI, machine learning offers a wealth of opportunities for

innovation and research. Whether you aim to build intelligent systems, analyze complex data, or

develop real-world applications, ML provides the tools and techniques to bring those ideas to

life.

Python

Python is the most widely used language in the field of data science and machine learning due

to its simplicity, readability, and strong community support. It offers a vast collection of libraries

like NumPy, Pandas, and scikit-learn that make it easy to manipulate data, train models, and

build applications. If you’re new to ML, mastering Python is the first step.

Numpy

NumPy (Numerical Python) is a core library for scientific computing in Python. It provides

support for large, multi-dimensional arrays and matrices, along with a collection of

mathematical functions to operate on them efficiently. Most ML workflows involve numerical

computations, making NumPy a foundational tool for model development and data manipulation.

Pandas

Pandas simplifies data analysis by offering powerful, user-friendly data structures like Series

and DataFrames. It allows you to load, clean, transform, and explore datasets with ease—tasks

that are essential for building reliable ML models. Mastering Pandas helps you handle real-world

data more efficiently.

Data Visualization

Visualizing your data helps you uncover hidden trends, relationships, and anomalies. Libraries

like Matplotlib and Seaborn allow you to create plots and charts that make your data more

understandable. Data visualization is also crucial for presenting findings and building

interpretable ML pipelines.

Data Preprocessing

Before training any model, the raw data must be cleaned and prepared. This involves handling

missing values, encoding categorical features, scaling numerical values, and selecting relevant

features. Proper Preprocessing ensures that models learn effectively and perform well on

unseen data.

Linear Regression

Linear regression is one of the simplest and most commonly used algorithms in supervised

learning. It models the linear relationship between one or more input features and a continuous

target variable. It forms the basis for more complex regression and time series models, making it

an ideal starting point.

Logistic Regression

Logistic regression is used for binary classification problems (e.g., spam or not spam). Despite

its name, it's a classification algorithm that uses a logistic function to estimate the probability

of a class. It’s interpretable, easy to implement, and works well for baseline models.


Decision Trees and Random Forests

Decision Trees split data into subsets based on feature values, creating a flowchart-like

structure that is easy to understand and interpret. Random Forest is an ensemble of decision

trees that improves prediction accuracy and reduces overfitting. These models are great for

both classification and regression tasks and are widely used in industry.

You can explore this section for the implementation: Decision Trees | Random Forest
Machine Learning Resources

SHAP (SHapley Additive exPlanations)

SHAP is a powerful model explainability technique based on cooperative game theory. It helps

attribute how much each feature contributes to a specific prediction, making even complex

models like Random Forests and XGBoost interpretable. SHAP is particularly useful in high-

stakes domains like healthcare or finance where transparency is key.

Optional Resources

Linear and Logistic Regression:

Linear Regression minimizes Mean Squared Error to fit a line to continuous data, while Logistic

Regression employs the sigmoid function to estimate probabilities for classification tasks. For a

more in-depth understanding of Linear and Logistic Regression, refer to this playlist.

Naive Bayes:

Naive Bayes is a probabilistic classifier based on Bayes' Theorem, with the assumption that

features are conditionally independent. It’s particularly effective for text classification tasks

like spam detection and sentiment analysis, due to its simplicity and performance on high-

dimensional data.

K-Nearest Neighbors:

KNN is an intuitive, non-parametric algorithm used for both classification and regression. It

works by comparing new data points to the ‘k’ closest training examples and predicting the

majority class (or average in regression). It’s easy to implement and understand, making it a

great choice for beginners.

Ev lu
a ation M etrics & Hyp p
er ara m eters:

Evaluating model performance requires the right metrics —mean squared error for Regression ,

1
or accuracy, precision, recall, and F -score for Classification
. Additionally, tuning

hyperparameters (like learning rate or tree depth) can significantly improve model

performance. Tools like GridSearchCV help automate this process.

Scikit-learn:

Scikit-learn is the most popular machine learning library in P ython. It provides tools for every

stage of an ML work fl
ow, from preprocessing and model training to evaluation and deployment

using a consistent and simple AP I.

Done with all the resources of Machine Learning?

Here’s something u can explore in the field of Deep Learning.

What is an LLM (Large Language Model)?

An LLM is an AI model trained on vast amounts of text to understand and generate human-like

language. It's capable of tasks like answering questions, writing content, translating languages,

and coding—all with remarkable fluency.

Why are LLMs special?

They generalize across tasks, generate coherent text, and need little or no task-specific

training (zero/few-shot learning). This versatility comes from the powerful Transformer

architecture, which uses self-attention to process and understand language context in parallel.

Revolution in AI: Transformers

Transformers replaced older sequential models (like RNNs) with parallel attention mechanisms,

enabling better performance and scalability. This is the core innovation behind models like GPT,

BERT, and others.

To know more about the transformer architecture, here’s a video to understand it.

What’s Next?

The task for the Machine Learning module will be released on 19th May. Please atleast go

through the Mandatory resources thoroughly without skipping anything .

Join the WhatsApp Group

Task Release - 19th May

Have any doubts ?

C ontact us

Prakha r
+ 91 9 3054 5358

1
C oding C lub

Shirshendu + 91 9 6743 7828