Machine Learning Basics For A Newbie


Machine learning is a vast and rapidly evolving field, and this article serves as a stepping stone for those new to the domain. Explore the fundamental concepts of machine learning, from understanding the differences between traditional programming and ML to delving into various types of machine learning algorithms.

Machine learning (ML) is a transformative field of artificial intelligence (AI) that empowers computers to learn and improve from experience without being explicitly programmed. Unlike traditional programming, where specific instructions are given to perform a task, ML algorithms enable systems to automatically learn patterns and make intelligent decisions based on data.

The significance of machine learning spans a multitude of industries, transforming the way we approach problem-solving and decision-making. Consider the following real-world examples.


  • Medical diagnosis: ML algorithms analyse patient data to assist in diagnosing diseases such as cancer and predicting patient outcomes.
  • Drug discovery: ML accelerates the drug discovery process by identifying potential candidates for new medications.


  • Fraud detection: ML models identify unusual patterns and detect fraudulent activities in financial transactions, safeguarding against unauthorised transactions.
  • Stock market prediction: Predictive models analyse historical data to forecast stock prices and optimise investment strategies.


  • Recommendation systems: ML algorithms power recommendation engines on platforms like Netflix and Amazon, providing personalised suggestions based on user preferences.
  • Customer segmentation: Businesses use ML to segment customers based on behaviour, enabling targeted marketing campaigns.


  • Speech recognition: Virtual assistants like Siri and Google Assistant utilise ML for accurate speech recognition and natural language processing.
  • Image recognition: ML enables systems to identify objects and patterns in images, contributing to facial recognition technology and security systems.

Understanding the basics of machine learning

Traditional programming vs machine learning: Traditional programming involves explicit instructions to perform a task, where developers define the rules and logic for the computer to follow. In contrast, machine learning relies on algorithms that allow systems to learn and adapt without being explicitly programmed. This adaptability makes ML well-suited for tasks where rules may be complex or unknown.

Role of data in ML: At the core of machine learning is data. ML algorithms learn from historical data, identifying patterns and relationships to make predictions or decisions. The quality and quantity of data directly impact the performance of ML models, highlighting the importance of collecting, cleaning, and preparing data for effective machine learning applications.

Types of machine learning

Supervised learning

Supervised learning involves training a model on a labelled dataset, where the algorithm learns the relationship between input features and corresponding output labels. Common examples include:

  • Image classification: Identifying objects in images.
  • Spam detection: Differentiating between spam and non-spam emails.

How it works: During training, the algorithm adjusts its parameters to minimise the difference between predicted and actual labels. The trained model can then make predictions on new, unseen data.

Unsupervised learning

Unsupervised learning deals with unlabelled data, aiming to discover patterns or relationships without predefined outputs. Examples include:

  • Clustering: Grouping similar data points together.
  • Dimensionality reduction: Simplifying complex datasets while retaining essential information.

How it works: Algorithms in unsupervised learning explore the inherent structure within the data, revealing hidden patterns or relationships without explicit guidance.

Reinforcement learning

Reinforcement learning involves an agent learning through trial and error, receiving feedback in the form of rewards or penalties. Examples include:

  • Game-playing AI: Learning optimal strategies by playing games.
  • Robotic control: Teaching robots to perform tasks through continuous feedback.

How it works: The agent takes actions in an environment, receives feedback, and adjusts its strategy to maximise cumulative rewards over time.

Key concepts in machine learning

Features and labels: In the context of machine learning, features are the measurable properties or characteristics of the data. For example, in an image recognition task, features might include pixel values, colours, and shapes. Choosing relevant features is crucial as they directly impact the model’s ability to learn and make accurate predictions.

Explanation of labels/targets: Labels, also known as targets or outputs, represent the desired outcome of the learning process. In a supervised learning scenario, the algorithm aims to predict these labels based on input features. For instance, in a spam detection model, the labels would indicate whether an email is spam or not.

Training data and testing data

Importance of splitting data: To evaluate the performance of a machine learning model, it’s essential to split the available data into two sets — training data and testing data. The model learns patterns from the training data, and the testing data assesses its ability to generalise to new, unseen examples. This division ensures a fair evaluation of the model’s performance.

Overfitting and underfitting: Overfitting and underfitting are common challenges in ML.

  • Overfitting occurs when a model learns the training data too well, including noise or irrelevant patterns, resulting in poor performance on new data.
  • Underfitting happens when a model is too simplistic and fails to capture the underlying patterns in the data, leading to poor performance overall.

Balancing between overfitting and underfitting is a crucial aspect of building effective ML models.

Common ML algorithms

Linear regression

Concept and use cases: Linear regression is a foundational supervised learning algorithm used for predicting a continuous outcome based on one or more input features. The model assumes a linear relationship between the input features and the target variable. Use cases include:

  • Predicting house prices: This is based on features like square footage, number of bedrooms, and location.
  • Revenue prediction: This involves estimating sales revenue based on marketing expenses.

How it works: Linear regression aims to find the best-fitting line (linear equation) that minimises the difference between predicted and actual values. This line is determined by adjusting weights assigned to each input feature during the training process.

Decision trees

Concept and use cases: Decision trees are versatile algorithms used for both classification and regression tasks. They represent decisions as a tree structure, with each node representing a decision based on a specific feature. Use cases include:

  • Credit scoring: Determining creditworthiness based on factors like income, debt, and credit history.
  • Medical diagnosis: Classifying diseases based on patient symptoms.

How it works: Decision trees split the dataset based on features to create subsets that are as pure as possible with respect to the target variable. This recursive process results in a tree structure, allowing the model to make decisions based on input features.

k-Nearest Neighbors (k-NN)

Concept and use cases: k-Nearest Neighbors is a simple yet effective algorithm for both classification and regression. It classifies or predicts a data point’s label based on the majority class or average of its k-nearest neighbours in the feature space. Use cases include:

  • Image recognition: Identifying similar images based on features.
  • Recommendation systems: Recommending items based on the preferences of similar users.

How it works: For a given data point, k-NN identifies the k-nearest neighbours in the feature space and assigns the majority class or average value as the prediction.

Resources for learning ML
Online courses and tutorials

  • Coursera
    Machine Learning by Andrew Ng: A popular course covering fundamental concepts and practical applications.
  • edX
    Introduction to Artificial Intelligence (AI) by Microsoft: Provides a broad overview of AI and its applications.
  • Kaggle courses
    Machine Learning Explainability: Helps understand and interpret machine learning models.

Books and recommended readings

  • ‘Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow’ by Aurélien Géron An excellent resource for practical hands-on experience with popular machine learning libraries.
  • ‘Python Machine Learning’ by Sebastian Raschka and Vahid Mirjalili
    A comprehensive guide using Python for machine learning and data science.

Hands-on projects and challenges

  • Kaggle
    Participate in Kaggle competitions to apply machine learning concepts to real-world problems.
  • GitHub repositories
    Explore machine learning projects and repositories on GitHub to learn from others and contribute to open source projects.

Machine learning workflow

Data collection and preparation

Importance of clean data: Clean and high-quality data is the foundation of successful machine learning. Data must be free of errors, inconsistencies, and missing values to ensure accurate model training. Data cleaning involves tasks such as handling outliers, addressing missing values, and transforming data into a suitable format.

Feature engineering: Feature engineering is the process of selecting, transforming, or creating features to enhance the performance of machine learning models. It involves identifying relevant features, scaling them appropriately, and creating new features that may improve the model’s predictive power.

Model training

Selection of algorithm: Choosing the right algorithm is a crucial step in the machine learning workflow. Consider the nature of the problem (classification, regression, etc) and the characteristics of the data when selecting an algorithm. Experimenting with different algorithms allows you to find the one that best fits your specific use case.

Tuning hyperparameters: Hyperparameters are parameters that are not learned during training but need to be set before training begins. Tuning hyperparameters involves adjusting these settings to optimise the performance of the model. Techniques such as grid search or random search help identify the best combination of hyperparameter values.

Model evaluation

Metrics for evaluation: Different machine learning tasks require different evaluation metrics. Common metrics include accuracy, precision, recall, and F1 score for classification tasks, while mean squared error or R-squared are often used for regression tasks. Understanding these metrics is essential for assessing how well a model is performing.

Cross-validation: Cross-validation is a technique used to assess a model’s performance by dividing the dataset into multiple subsets. The model is trained and evaluated on different subsets, ensuring that it generalises well to new, unseen data.

Model deployment

Making predictions on new data: Once a model has been trained and evaluated, it is ready for deployment. Deployment involves using the model to make predictions on new, real-world data. This could be in the form of a web application, an API, or integration into existing systems.

Real-world applications: Machine learning models find application in various real-world scenarios, from predicting customer behaviour in e-commerce to optimising supply chain logistics. The successful deployment of a model contributes to solving practical problems and making informed decisions in diverse domains.

Challenges and ethical considerations in machine learning

Bias in machine learning

Understanding bias: Bias in machine learning refers to the presence of systematic errors in a model’s predictions that result from the training data. If the training data is not representative or contains biases, the model may perpetuate and even amplify those biases, leading to unfair or discriminatory outcomes.

Ways to address bias: Addressing bias in machine learning requires careful consideration and proactive measures:

  • Diverse and representative data: Ensure that training data is diverse and representative of the population it aims to serve.
  • Bias detection and mitigation: Use techniques to identify and mitigate bias, such as re-sampling, re-weighting, or incorporating fairness constraints into the learning process.

Privacy concerns

Data privacy in ML: Machine learning often involves the use of sensitive data, raising concerns about privacy. It is essential to safeguard individuals’ privacy by implementing measures such as anonymization, encryption, and secure data storage.

Compliance with regulations: Adherence to data protection regulations, such as GDPR, HIPAA, or other industry-specific standards, is crucial. ML practitioners must be aware of and comply with legal and ethical frameworks to protect user privacy.

Whether you’re a student embarking on a machine learning course, a professional seeking to integrate ML into your work, or an enthusiast eager to understand the technology shaping our future, the world of machine learning offers endless possibilities for learning and innovation. Remember, the best way to solidify your understanding is through hands-on practice. Experiment with different datasets, work on projects, and engage with the vibrant machine learning community to stay informed and inspired. As you continue your journey, consider diving deeper into specific algorithms, exploring advanced topics like deep learning, and applying your knowledge to real-world projects.


Please enter your comment!
Please enter your name here