Course Details

Courses Descriptions

MODULE OBJECTIVE

What you'll achieve

Apply statistical exploration to real data, refresh the maths underpinning ML/DL, and understand and utilize core supervised, unsupervised, and heuristic learning algorithms.

SECTION 01 — Univariate Analysis

Single-variable exploration

• Categorical: Count, Count%, Pie chart, Bar chart
• Numerical: Min, Max, Mean, Median, Mode
• Spread: Range, Quartiles, Variance, Std Dev, CV
• Shape: Skewness, Kurtosis — Histogram, Box plot
• Encoding (Cat→Num) and Binning (Num→Cat)

SECTION 02 — Bivariate Analysis

Relationships between two variables

• Cat vs Cat: Chi-squared test, Bar / 2-Y axis plot
• Num vs Num: Correlation, Scatter plot
• Cat vs Num: Z test, t test, ANOVA
• Correlation does not imply causation

SECTION 03 — Maths for ML & DL

Linear algebra, probability, calculus

• Vectors, matrices, dot products, norms
• Transpose, inverse, eigenvalues / eigenvectors
• Probability: Bayes theorem, distributions
• Calculus: derivatives, chain rule, gradients
• Gradient descent intuition

Implement matrix multiplication manually, Visualize vectors using Python

SECTION 04 — Supervised ML

Classification and regression

• Frequency Table: ZeroR, OneR, Naive Bayes, Decision Tree
• Covariance Matrix: LDA, Logistic Regression, MLR
• Similarity: K Nearest Neighbors
• Other: ANN, Support Vector Machine
• Evaluation: Confusion matrix, F1, AOC, RMSE

Predict House Prices. Predict Salaries. Forecast Sales. Diabetes Prediction. Fraud Detection.

SECTION 05 — Unsupervised & Association Rules

Clustering and pattern discovery

• Hierarchical: Agglomerative, Divisive
• Partitive: K-Means, Self Organising Map
• Elbow method and Silhouette score
• Apriori: Support, Confidence, Lift

KEY TAKEAWAYS

1. Exploration must always precede modelling

2. Visualise data before fitting any model

3. Gradient descent is the engine of all deep learning

4. Unsupervised methods uncover structure without labels