Applied Classical Machine Learning

This course turns you from an ML beginner into a practitioner, teaching you to build deployable solutions using Scikit-learn and XGBoost. Through hands-on labs like predicting diabetes progression and detecting cancer, you'll master key algorithms and model tuning. Learn to diagnose overfitting and deliver actionable insights for real-world challenges.

Instructor: Ridwan Ibidunni

Available to members only

Prerequisite

  • Python + Pandas/NumPy

  • Data preprocessing

  • Data visualization

  • Basic stats

  • Google Colab

  • Recommended: Completion of “Python for ML”

Modules & Duration

  • 8 modules

  • 7 weeks + final project

Assignments

  • 14 assignments

  • 4 projects

What you'll learn

Foundations of ML: Differentiate supervised/unsupervised learning and master bias-variance tradeoffs.

Ensemble Methods: Boost accuracy with Random Forests and XGBoost (e.g., Titanic survival prediction).

Regression Mastery: Implement linear/polynomial regression, regularization (Ridge/Lasso), and predict real-world trends (e.g., COVID-19 cases)

Unsupervised Learning: Cluster customers using k-means, reduce dimensions with PCA/t-SNE, and extract business insights. 

Classification Techniques: Build logistic regression, SVM, and decision tree models for problems like cancer detection and spam classification. 

Model Evaluation: Diagnose overfitting, optimize hyperparameters via GridSearchCV, and select metrics (ROC-AUC, F1-score). 

Skills you’ll gain

Data splitting

Precision

recall

PCA

Decision trees

Random Forest

t-SNE

Supervised vs. unsupervised learning

SVM

AdaBoost

F1-score

ROC curve

Scikit-learn

Bias-variance tradeoff

K-means

cross-validation

XGBoost

Matplotlib

Hierarchical clustering

hyperparameter tuning

Add certificate to your LinkedIn profile

Taught in English

Regularization (Ridge, Lasso)

Google Colab

VS Code

Linear/polynomial regression

Logistic regression

Machine learning drives decisions in healthcare, finance, and tech; but only when theory meets practice. This course transforms you from an ML novice into a practitioner who builds deployable solutions. Through hands-on labs; like predicting diabetes progression, detecting cancer, and segmenting retail customers, you’ll master classical algorithms using Scikit-learn and XGBoost. Learn to diagnose overfitting, tune models for real-world noise, and deliver insights that stakeholders trust. Enroll to turn data into decisions. 

Brief Course Description - 8 modules

  • What you'll learn:

    Identify ML types, split datasets, and diagnose bias-variance tradeoffs. 
    Includes: 

    • 2 hours of live virtual lectures 

    • Hands-on: Polynomial overfitting lab, data splitting with stratification 

    • Assignment: Analyze a real-world ML use case (e.g., recommendation systems) 

    • Dataset: Iris 

  • What you'll learn:

    Implement linear regression, interpret coefficients, and handle skewed data. 
    Includes: 

    • 2 hours of live virtual lectures 

    • Hands-on: Boston Housing price prediction, residual analysis 

    • Debug Challenge: Fix skewed target variables 

    • Tools: Scikit-learn, Seaborn 

  • What you'll learn:

    Combat overfitting with Ridge/Lasso regression and balance L1/L2 penalties. 
    Includes: 

    • 2 hours of live virtual lectures 

    • Hands-on: Diabetes dataset regularization, coefficient shrinkage 

    • Project: Predict COVID-19 case trends 

    • Dataset: Diabetes, COVID-19 

  • What you'll learn:

    Build binary classifiers, tune probability thresholds, and visualize decision boundaries. 
    Includes: 

    • 2 hours of live virtual lectures 

    • Hands-on: Spam email classification, SVM kernel comparison 

    • Debug Challenge: Address class imbalance with SMOTE 

    • Dataset: Email Spam 

  • What you'll learn:

    Train interpretable trees, evaluate models using confusion matrices, and optimize precision-recall. 
    Includes: 

    • 2 hours of live virtual lectures 

    • Hands-on: Iris species classification, ROC curve plotting 

    • Project: Cancer detection with Wisconsin dataset 

    • Tools: Scikit-learn

  • What you'll learn:

    Leverage Random Forests for robustness and XGBoost for speed/accuracy tradeoffs. 
    Includes: 

    • 2 hours of live virtual lectures 

    • Hands-on: Customer churn prediction, feature importance analysis 

    • Assignment: Titanic survival prediction 

    • Dataset: Titanic 

  • What you'll learn:

    Cluster customers via k-means/hierarchical methods and compress dimensions with PCA/t-SNE. 
    Includes: 

    • 2 hours of live virtual lectures 

    • Hands-on: RFM-based customer segmentation, MNIST visualization 

    • Project: Retail customer profiling 

    • Tools: Scikit-learn, Seaborn 

  • What you'll learn:

    Optimize models using cross-validation and evaluate imbalanced data. Includes: 

    • Guided project: End-to-end Kaggle competition 

    • Tasks: Fraud detection with precision-recall/ROC analysis 

    • Deliverable: Business-ready ML pipeline 

    • Grading: Functionality (40%), EDA (20%), tuning (10%) 

Why This Course Stands Out 

Industry Projects: Solve problems with healthcare (cancer detection), finance (fraud), and retail (segmentation) datasets.

Debug-First Approach: Fix real-world issues like overfitting, leakage, and imbalance in every module.

Peer-Driven Learning: Collaborative reviews and pair programming mirror workplace workflows. 

Portfolio: Showcase 4 projects on LinkedIn. 

Previous

Python for Machine Learning