Topic outline

  • Course Introduction

    Collapse all Expand all
    • Time: 19 hours
    • CEUs: 1.9
    • Free Certificate

    • Enroll Me in This Course
      Confirm on the next page

    • This comprehensive course is designed to equip you with a strong foundation in machine learning (ML) through a systematic, step-by-step approach. This course covers the essential principles of supervised and unsupervised learning algorithms, providing a deep understanding of how machine learning models work and how they can be applied in real-world scenarios. You will explore the entire ML workflow, from data collection and preprocessing to model building and evaluation, ensuring you gain practical, hands-on experience at each stage.

      Throughout the course, you will master key concepts in data preprocessing, feature engineering, and model evaluation techniques. We will cover a range of core algorithms, including regression, classification, and clustering, as well as evaluation metrics to help you assess model performance and make data-driven decisions. Practical exercises and Python-based implementations will reinforce your understanding and allow you to build predictive models. By the end of the course, you will be equipped to handle complete machine learning projects, from data preparation to evaluation, while ensuring your models are both effective and ethical.

      In addition to the technical skills, this course emphasizes the importance of ethical decision-making in AI development. You will explore critical issues like bias, fairness, and accountability in machine learning, learning how to build models that are not only accurate but also responsible and equitable. Whether you want to enhance your career, pursue further studies, or contribute to the growing field of AI, CS207 provides you with the knowledge and skills necessary to create impactful and ethical machine learning systems.

    • page icon
      Course Introduction Video Page
      Students must
      View
    • page icon
      Course Syllabus Page

  • Unit 1: Introduction to Machine Learning

    Welcome to the exciting world of machine learning! This unit is the gateway to your journey into one of today's most dynamic and rapidly evolving technological fields. Machine learning (ML) is transforming industries and shaping our daily lives, from the personalized recommendations you see on streaming platforms to the cutting-edge self-driving cars of the future. Its profound impact is felt in sectors ranging from health care to finance, helping solve complex problems and driving innovation.

    In this unit, we will start by defining machine learning and why it is important in the modern technological landscape. We will then explore its relationship with artificial intelligence (AI) and data science, placing ML within these broader contexts and illustrating how these fields work together to create intelligent, data-driven systems.

    You will also learn about the three core types of machine learning – supervised, unsupervised, and reinforcement learning. We will examine the unique characteristics and applications of each, providing real-world examples to illustrate how these techniques are applied in various domains. Understanding these distinctions will equip you with the knowledge to choose the right machine learning approach for different problems. By the end of this unit, You will have a solid foundation in machine learning, recognizing its versatility and impact in tackling diverse challenges across industries.

    Completing this unit should take you approximately 1 hour.

    • 1.1: What is Machine Learning?

      In this section, we will explore the formal definition of ML and how it differs from traditional rule-based programming. This will include how machines ""learn"" from data and the role of algorithms in this process. We will explore real-world applications of ML across various industries.

      By the end of this section, you will have a strong foundational understanding of ML and its impact on various domains. As we progress, please take note of real-world examples – this will help you develop a deeper and more practical understanding of the material.

    • 1.2: Types of Machine Learning

      In this section, you will explore the three main types of ML, how they differ in their learning process, and the strengths and limitations of each kind in solving different problems. We will explore real-world applications of each ML type. By the end of this section, you will be able to distinguish between different ML types, understand their applications, and analyze their relevance in real-world scenarios. As you progress, keep a list of examples you encounter that use supervised, unsupervised, or reinforcement learning -- this will help you build a deeper and more practical understanding of the material.

    • 1.3: ML vs. AI vs. Data Science

      This subunit will explore the definitions of AI, ML, and data science and their key differences. This includes how ML fits within AI, why it is considered a subset, and the role of data science in working with large datasets and deriving insights. We will explore real-world applications where these fields intersect, such as healthcare, finance, and autonomous systems. It is important to understand how these fields work together. For example, in healthcare, AI can diagnose diseases, ML improves diagnostic accuracy, and data science helps process and analyze vast amounts of patient data.

      By the end of this section, you should be able to clearly differentiate between AI, ML, and data science and understand how they complement each other in advancing modern technology. As you progress, keep track of real-world examples that use these technologies – this will help you develop a deeper, more practical understanding of the material.

    • Unit 1 Assessment

  • Unit 2: Machine Learning Workflow

    Building a successful machine learning model involves more than selecting the right algorithm. It requires following a systematic and well-defined machine learning workflow or pipeline process. This unit will guide you through the essential stages of the machine learning process, from the very beginning – data collection and preparation – to the final steps of model evaluation and deployment.

    Understanding each stage of the workflow is critical for the success of your machine learning projects. We will explore best practices for data collection, techniques for effective data preprocessing, strategies for selecting and training the right models, and methods for evaluating the performance of your models. Each step is vital in ensuring that your model is accurate, robust, and ready for real-world applications.

    By the end of this unit, you will have a comprehensive understanding of the machine learning pipeline. You will be well-equipped to confidently approach any ML project. Whether You are working with small datasets or large-scale problems, this foundational knowledge will provide you with the tools to build effective and deployable machine learning models.

    Completing this unit should take you approximately 1 hour.

    • 2.1: The Machine Learning Pipeline

      In this section, we will explore the stages of the machine learning pipeline. As you work through this section, consider these questions. How do different stages in the ML pipeline contribute to the model's overall effectiveness? Why is preprocessing critical before training a model? What role does feature engineering play in improving model performance? How does monitoring ensure long-term success for ML models? What challenges might arise in deploying ML models in real-world scenarios?

    • 2.2: Significance of Each Stage

      In this section, we will explore the significance of each stage in the machine learning pipeline. As you work through this section, consider these questions. Why is defining the problem important before starting an ML project? How does data quality impact model performance? What role does feature engineering play in improving predictions? How does model evaluation help ensure reliability? What are the challenges of deploying and maintaining ML models?

    • Unit 2 Assessment

  • Unit 3: Data Preprocessing

    The quality of your data is paramount to the success of your model. Raw data, in its natural form, is often messy, incomplete, or inconsistent, hindering the learning process and leading to inaccurate predictions. This unit focuses on the important step of data preprocessing, where we transform raw data into a format suitable for machine learning algorithms to work with effectively.

    You will learn various techniques to clean your data, addressing common issues such as missing values, noise, and outliers. Data preprocessing is not just about cleaning – it also involves transforming and normalizing your data to ensure consistency and fairness. Techniques like scaling and standardization help bring features to a similar scale, preventing any one feature from dominating the learning process. We will also cover methods for encoding categorical variables and turning non-numerical data into a form machine learning algorithms can understand and process.

    Data preprocessing is the foundation of any successful machine learning project. Mastering these essential techniques can transform raw data into high-quality datasets, leading to accurate, reliable, and robust machine learning models. This unit will equip you with the tools to handle data effectively and prepare it for use in predictive modeling.

    Completing this unit should take you approximately 2 hours.

    • 3.1: Data Cleaning Techniques

      In this section, we will consider data cleaning techniques. By applying structured data cleaning methods, you can enhance the quality of machine learning datasets, improving model performance and reliability. As you review this section, focus on how each technique contributes to preparing high-quality data for machine learning workflows.

    • 3.2: Normalization

      In this section, we will explore the concept of normalization. By the end of this section, you should be able to recognize the importance of normalization in machine learning models, identify when to use different normalization techniques, and apply appropriate normalization methods to improve model performance.

    • 3.3: Data Transformation: Encoding Categorical Variables

      Understanding categorical data is essential in machine learning, as models require numerical inputs rather than raw text categories. This section introduces different methods for encoding categorical data, helping you develop a solid foundation for effectively preparing data. Kindly pay close attention to vocabulary encoding and one-hot encoding. Additionally, this section discusses handling outliers in categorical data by grouping rare values into an "out-of-vocabulary" category.

    • Unit 3 Assessment

  • Unit 4: Data Visualization

    Data visualization is essential for anyone working in data science and machine learning. It allows us to see the stories hidden within data, recognize patterns, and communicate insights effectively. This unit explores visualization techniques and their applications in machine learning, including scatter plots, histograms, box plots, and heat maps. You will learn when and how to use each type of plot to convey meaningful information about your dataset.

    Interpreting visual data is a critical aspect of the machine learning process. We will focus on extracting key insights from these visualizations, such as identifying correlations, detecting trends, and spotting anomalies that may impact model performance. Additionally, we will introduce feature engineering – creating new features from existing ones to improve your models. Visualization plays an important role in this process by helping you identify potential features and evaluate their importance. By the end of this unit, you will have a solid understanding of how to effectively visualize and interpret data and how these visual techniques can enhance your machine learning workflow and decision-making.

    Completing this unit should take you approximately 2 hours.

    • 4.1: Data Visualization Techniques

      Effective data visualization allows us to explore datasets, identify patterns, and communicate insights clearly. In this section, you will learn how to create scatter plots, histograms, box plots, bar charts, and heatmaps using Python libraries like Matplotlib and Seaborn. By the end of this section, you should be able to choose the appropriate visualization for different types of data and confidently use Python libraries to generate meaningful plots that enhance data analysis and decision-making.

    • 4.2: Interpreting Visual Data

      Data visualization is a powerful tool for uncovering insights that might not be immediately apparent in raw data. In this section, you will explore techniques for identifying patterns, trends, and anomalies in visual representations of data. Recognizing these elements is important for making informed decisions, detecting relationships between variables, and spotting unusual data points that could indicate errors or significant findings.

    • 4.3: Feature Engineering

      In this section, you will explore how to create meaningful features from existing data and select the most relevant ones to improve model performance. By the end of this section, you should understand how feature engineering enhances model performance and be able to apply these techniques effectively in machine learning tasks.

    • Unit 4 Assessment

  • Unit 5: Supervised Learning – Regression

    In this unit, we discuss supervised learning, a powerful technique where models are trained on labeled data to make predictions. Specifically, we will focus on regression, a type of supervised learning used to predict continuous values. We will start with simple linear regression, a fundamental method for modeling the relationship between two variables and predicting outcomes based on linear patterns.

    Building on this foundation, we will then explore multiple linear regression, which allows us to model more complex relationships involving multiple predictor variables. This technique is widely used in real-world scenarios where multiple factors influence the outcome. Finally, we will explore the evaluation of regression models, covering key metrics such as R-squared and Root Mean Squared Error (RMSE). We will also discuss the limitations of linear regression, including the important concept of multicollinearity, which can impact model accuracy.

    By the end of this unit, You will have a strong understanding of regression techniques, the ability to implement them in Python, and the knowledge to evaluate and apply these models effectively in real-world applications.

    Completing this unit should take you approximately 2 hours.

    • 5.1: Introduction to Regression

      Linear regression is a fundamental technique for modeling relationships between variables. This section explores key concepts using Python and Scikit-Learn, with practical examples from the given resource.

    • 5.2: Evaluating Regression Models

      Evaluating a regression model is essential to ensure its predictive accuracy and reliability. Key metrics like MAE, MSE, RMSE, R², Adjusted R², and Cross-validated R² provide insights into model performance. MAE measures the average error magnitude, MSE and RMSE penalize larger errors, R² explains the proportion of variance in the target variable, Adjusted R² accounts for model complexity by penalizing unnecessary predictors, and Cross-validated R² assesses generalization performance. Together, these metrics form a comprehensive framework for refining and selecting the best regression model. This section explores these metrics to evaluate regression model performance, determining predictive accuracy and reliability.

    • 5.3: Limitations of Regression Models

      In this section, you will examine common challenges in regression analysis, such as multicollinearity, extrapolation, and other factors influencing model reliability. These resources will help you recognize and address these issues effectively.

      As you work through this section, consider these questions. Are there outliers or influential observations affecting your model? Does your data follow a linear pattern, or should you apply transformations?

    • Unit 5 Assessment

  • Unit 6: Supervised Learning – Classification

    In this unit, we continue with supervised learning by focusing on classification, a technique used to predict categorical outcomes. Classification is crucial in many machine learning applications, such as spam detection, medical diagnoses, and image recognition. We will begin with logistic regression, one of the most widely used methods for binary classification problems, and explore its implementation using Python.

    You will learn how to build a logistic regression model, interpret its coefficients, and make predictions. Evaluating the model's performance is a key part of classification, and we will cover important metrics such as accuracy, precision, recall, and F1-score. By the end of this unit, you will have the knowledge and skills to implement, evaluate, and refine classification models for a wide range of practical problems, ensuring you can apply these techniques effectively in real-world scenarios.

    Completing this unit should take you approximately 3 hours.

    • 6.1: Logistic Regression

      This section will explore logistic regression, a fundamental classification algorithm used to predict the probability of an event occurring. Unlike linear regression, which produces continuous outputs, logistic regression applies the sigmoid function to transform predictions into probability values between 0 and 1. This makes it useful for tasks such as spam detection, medical diagnosis, and fraud detection.

    • 6.2: Classification

      This section will explore classification, how it works, and how thresholding affects model performance in different contexts.

    • 6.3: Evaluating Classification Models

      This section evaluates classification models and the importance of accuracy, precision, recall, and ROC-AUC. By exploring these metrics, you will learn how to evaluate classification models effectively and make informed decisions based on the problem at hand.

    • Unit 6 Assessment

  • Unit 7: Unsupervised Learning – Clustering

    In this unit, we transition to unsupervised learning, where we work with unlabeled data to uncover hidden patterns and structures. Clustering is one of the most powerful techniques in unsupervised learning, and it is used to group similar data points together. We will also discuss K-means clustering, a widely used and intuitive method for partitioning data into distinct clusters. You will learn how to implement K-means clustering using Python to determine the optimal number of clusters for a given dataset.

    A key part of the clustering process is analyzing and interpreting the results. We will explore methods for evaluating the quality of clusters and visualizing cluster assignments, which will help you gain deeper insights into the underlying structure of your data.

    Clustering has many practical applications, such as customer segmentation, anomaly detection, and data exploration. By the end of this unit, you will be equipped to implement and interpret clustering models, uncover hidden patterns in your data, and make informed, data-driven decisions.

    Completing this unit should take you approximately 3 hours.

    • 7.1: Introduction to Clustering

      This section focuses on clustering, an unsupervised learning technique that groups similar data points together based on their characteristics. Unlike classification, clustering does not rely on predefined labels. Instead, it identifies natural patterns within the data. By exploring these concepts, you will develop a strong foundation in clustering and learn how to apply it effectively to uncover hidden patterns in data.

    • 7.2: K-Means Clustering

      K-means clustering is a widely used unsupervised machine learning algorithm that groups data points based on similarity. It divides data into K distinct, non-overlapping clusters, assigning each point to the nearest cluster centroid (mean of the cluster). K-means is popular for customer segmentation, image compression, and anomaly detection due to its simplicity, efficiency, and scalability. However, it requires specifying the number of clusters (K) in advance and performs best with spherical, well-separated data. This section will explain the K-means clustering algorithm, its steps, assumptions, limitations, and strategies to improve its effectiveness.

    • 7.3: Analyzing Clustering Results

      This section explores the process of analyzing and evaluating K-means clustering results, focusing on similarity measures and clustering performance. These resources will help you understand cluster formation, interpret results effectively, and refine your models for improved performance.

    • Unit 7 Assessment

  • Unit 8: Model Evaluation and Validation

    In this unit, we focus on one of the most important aspects of machine learning: evaluating and validating models. Building robust, reliable models isn't just about training them well – it's also about ensuring they can generalize to unseen data. We will begin by discussing the train-test split, which divides data into training and testing sets to evaluate model performance on new, unseen data.

    We will also explore cross-validation, a more robust method for assessing model performance by training and testing the model on different subsets of the data. This technique provides a more reliable estimate of how the model will perform in real-world scenarios.

    A key challenge in machine learning is overfitting, where a model excels on training data but fails on new data. We will discuss how to recognize and address both overfitting and how techniques like regularization can help prevent these issues.

    By the end of this unit, you will have a deep understanding of the bias-variance tradeoff and be equipped with tools and strategies to evaluate and fine-tune your models, ensuring they perform consistently across various datasets.

    Completing this unit should take you approximately 2 hours.

    • 8.1: Train-Test Split and Cross-Validation

      This section introduces the importance of splitting data into training and testing sets to evaluate model performance effectively. It also covers cross-validation, a robust technique for assessing model generalizability by partitioning data into multiple subsets and iteratively training and testing the model. These methods help prevent overfitting and ensure reliable performance on unseen data.

    • 8.2: Overfitting and Underfitting

      Building an effective machine learning model requires balancing learning from data and generalizing it to new examples. This section will explore the concepts of fitting, overfitting, and underfitting – key considerations in developing models that perform well in real-world scenarios.

    • 8.3: Techniques to Avoid Overfitting

      Overfitting is a common challenge in machine learning. A model that overfits performs exceptionally well on training data but struggles to generalize to new, unseen data. In this section, you will explore key strategies to mitigate overfitting, such as managing model complexity, applying regularization techniques, and interpreting loss curves.

    • Unit 8 Assessment

  • Unit 9: Practical Implementation of ML Models

    In this unit, we combine all the concepts and skills you have learned throughout the course and focus on the practical implementation of machine learning models in real-world projects. You will learn how to develop an end-to-end machine learning project, covering the entire process – from data collection and preprocessing to model implementation.

    A key aspect of any machine learning project is ensuring documentation and reproducibility. You will discover best practices for documenting your workflow, using version control systems like Git, and creating code that can be easily reproduced. By the end of this unit, you will have hands-on experience developing and presenting an ML project, preparing you to confidently tackle real-world machine learning challenges. This unit will equip you with the practical skills to move from theory to implementation, ensuring your machine learning models are effective and ready for deployment in real-world scenarios.

    Completing this unit should take you approximately 2 hours.

    • 9.1: Developing an ML Project

      This section provides a step-by-step guide to implementing logistic regression in Python. This section covers problem definition, logistic regression basics, data preparation, model building, and execution.

    • 9.2: Project Documentation and Reproducibility

      Reproducibility is a fundamental principle of scientific research that ensures others can consistently obtain results following the same methods. This section focuses on computational reproducibility, which involves documenting and structuring code, data, and workflows so that others can replicate the same analysis and arrive at identical results.

    • 9.3: Version Control

      This section introduces version control and Git, essential tools for tracking changes, collaborating on projects, and maintaining organized workflows.

    • Unit 9 Assessment

  • Unit 10: Ethical and Responsible AI

    Machine learning and artificial intelligence continue to shape our world, and addressing the ethical implications of these technologies is very important. In this unit, we will explore the key ethical considerations surrounding AI, such as bias in data and algorithms, fairness in model outcomes, privacy concerns, and the need for transparency and accountability in AI systems.

    Mitigating bias in machine learning models helps work towards creating fairer, more equitable systems. We also explore responsible AI practices, offering guidelines for developing and deploying ethical AI solutions that align with societal values. As data scientists, it is our duty to ensure that the technologies we create have a positive, inclusive impact on society. By the end of this unit, you will be equipped with the knowledge of ethical considerations in ML and contribute to developing technology that benefits everyone.

    Completing this unit should take you approximately 1 hour.

    • 10.1: Ethical Considerations in ML

      This section explores the five key pillars of ethical AI: Accountability, Reliability, Explainability, Security, and Privacy. It highlights the importance of integrating these principles into machine learning systems to ensure fairness, safety, and trust. Ethical AI practices can shape responsible innovation and build public trust in technology. This foundational knowledge will help you critically evaluate and implement ethical considerations in your own ML projects.

    • 10.2: Responsible AI Practices

      Responsible AI focuses on creating reliable, fair, and trustworthy systems. This involves ensuring model robustness, identifying and correcting label errors, and mitigating biases that can lead to unfair outcomes. Using tools like CheckList and Cleanlab and prioritizing ethical practices such as diverse data and continuous monitoring, we can build AI systems that perform accurately and equitably. Responsible AI is essential for building trust, reducing risks, and ensuring AI benefits society as a whole.

    • Unit 10 Assessment

  • Study Guide

    This study guide will help you get ready for the final exam. It discusses the key topics in each unit, walks through the learning outcomes, and lists important vocabulary. It is not meant to replace the course materials.

  • Certificate Final Exam

    Take this exam if you want to earn a free Course Completion Certificate.

    To receive a free Course Completion Certificate, you will need to earn a grade of 70% or higher on this final exam. Your grade for the exam will be calculated as soon as you complete it. If you do not pass the exam on your first try, you can take it again as many times as you want, with a 7-day waiting period between each attempt. Once you pass this final exam, you will be awarded a free Course Completion Certificate.

  • Course Feedback Survey

    Please take a few minutes to give us feedback about this course. We appreciate your feedback, whether you completed the whole course or even just a few resources. Your feedback will help us make our courses better, and we use your feedback each time we make updates to our courses. If you come across any urgent problems, email contact@saylor.org.