
Already know some of this? A 2-minute test skips the lessons you know.
Take the placement testA free 20-lesson course built from The iScale's 5h 28m of YouTube video. 4 days at 90 minutes a day, with study notes and a quiz in every lesson. A 2-minute placement test skips what you already know.
Taught by The iScale on YouTube. Videos play in YouTube's own player, so the creator keeps every view.
Understand the course structure, the relationship between AI, ML, and Deep Learning, and real-world applications like recommendation systems and fraud detection.
Learn the end-to-end ML model workflow (data preprocessing, training, evaluation) and distinguish between supervised, unsupervised, and reinforcement learning paradigms.
Differentiate regression (continuous output) from classification (categorical output), including binary and multi-class classification, and identify dependent vs independent features.
Derive the linear regression hypothesis, cost function (MSE), and gradient descent update rules to find the best-fit line for predicting continuous values.
Implement Linear Regression in Jupyter Notebook using the Boston Housing dataset: load data, split features/target, train model, predict, and evaluate with R² and RMSE.
Understand overfitting/underfitting, bias-variance tradeoff, and how Ridge (L2) and Lasso (L1) regularization prevent overfitting, with Lasso enabling automatic feature selection.
Implement Ridge and Lasso regression with GridSearchCV for hyperparameter tuning, compare their MSE scores against Linear Regression on the Boston Housing dataset.
Learn the sigmoid function, why linear regression's cost function fails for classification, the log-loss cost function, and evaluation metrics: Confusion Matrix, Precision, Recall, F1-Score.
Implement Logistic Regression on the Breast Cancer dataset: load data, split/scale features, train model, predict, and evaluate using confusion matrix heatmap and classification report.
Understand conditional probability, Bayes Theorem derivation, and the Gaussian Naive Bayes assumption for continuous features; learn when to use it over other classifiers.
Implement Gaussian Naive Bayes on the Iris dataset: import libraries, load data, train/test split, fit model, predict, and evaluate accuracy and confusion matrix.
Learn how KNN works for classification and regression using Euclidean and Manhattan distance, and understand its sensitivity to outliers and imbalanced data.
Understand Decision Tree structure (root, internal, leaf nodes), splitting criteria (Entropy, Gini Impurity), and feature selection via Information Gain for pure splits.
Implement Decision Tree Classifier on Iris dataset, visualize the tree structure with plot_tree, and interpret Gini impurity values at each node to understand splitting logic.
Learn Bagging (bootstrap aggregating) vs Boosting (sequential error correction), and implement Random Forest with feature importance visualization on the Breast Cancer dataset.
Understand AdaBoost's sequential training of weak learners (decision stumps), weight updating for misclassified points, and implement it on Breast Cancer data achieving 97% accuracy.
Learn K-Means iterative centroid optimization, use the Elbow Method to find optimal K, and visualize clusters on Iris dataset using 2D scatter plots with centroids.
Understand agglomerative clustering (bottom-up merging), dendrogram interpretation with Euclidean distance, and practical implementation using scipy's linkage and dendrogram.
Learn DBSCAN's core concepts (eps, min_samples, core/border/noise points) and implement it on synthetic moon-shaped data to discover arbitrary-shaped clusters.