Get in Touch
 Duration 21 hours

Course Outline

Introduction

This section offers an overview of when to apply 'machine learning', key considerations, and its implications, covering both advantages and limitations. It addresses data types (structured/unstructured/static/streamed), data validity and volume, data-driven versus user-driven analytics, statistical models versus machine learning models, the challenges of unsupervised learning, the bias-variance trade-off, iteration and evaluation, cross-validation techniques, and the distinctions between supervised, unsupervised, and reinforcement learning.

MAJOR TOPICS

1. Grasping the naive Bayes concept

  • Fundamental concepts of Bayesian methods
  • Probability theory
  • Joint probability
  • Conditional probability using Bayes' theorem
  • The naive Bayes algorithm
  • Classification via naive Bayes
  • The Laplace estimator
  • Handling numeric features with naive Bayes

2. Grasping the concept of decision trees

  • The divide and conquer approach
  • The C5.0 decision tree algorithm
  • Selecting the optimal split
  • Pruning decision trees

3. Grasping the concept of neural networks

  • Transitioning from biological to artificial neurons
  • Activation functions
  • Network topology
  • Determining the number of layers
  • Information flow direction
  • Determining node count per layer
  • Training neural networks using backpropagation
  • Deep Learning

4. Grasping the concept of Support Vector Machines

  • Classification using hyperplanes
  • Determining the maximum margin
  • Scenarios with linearly separable data
  • Scenarios with non-linearly separable data
  • Applying kernels to non-linear spaces

5. Grasping the concept of clustering

  • Clustering as a machine learning objective
  • The k-means clustering algorithm
  • Assigning and updating clusters using distance
  • Selecting the optimal number of clusters

6. Assessing classification performance

  • Working with classification prediction data
  • In-depth analysis of confusion matrices
  • Utilizing confusion matrices for performance measurement
  • Performance metrics beyond accuracy
  • The kappa statistic
  • Sensitivity and specificity
  • Precision and recall
  • The F-measure
  • Visualizing performance tradeoffs
  • ROC curves
  • Predicting future performance
  • The holdout method
  • Cross-validation
  • Bootstrap sampling

7. Optimizing standard models for improved performance

  • Leveraging caret for automated parameter tuning
  • Developing a basic tuned model
  • Customizing the tuning workflow
  • Enhancing model performance through meta-learning
  • Understanding ensembles
  • Bagging
  • Boosting
  • Random forests
  • Training random forests
  • Evaluating random forest performance

MINOR TOPICS

8. Understanding classification via nearest neighbors

  • The kNN algorithm
  • Distance calculation
  • Selecting an appropriate k
  • Data preparation for kNN
  • Why the kNN algorithm is considered lazy?

9. Understanding classification rules

  • The separate and conquer method
  • The One Rule algorithm
  • The RIPPER algorithm
  • Deriving rules from decision trees

10. Understanding regression

  • Simple linear regression
  • Ordinary least squares estimation
  • Correlations
  • Multiple linear regression

11. Understanding regression trees and model trees

  • Incorporating regression into trees

12. Understanding association rules

  • The Apriori algorithm for association rule mining
  • Measuring rule relevance – support and confidence
  • Constructing rule sets using the Apriori principle

Extras

  • Spark/PySpark/MLlib and Multi-armed bandits

Requirements

Proficiency in Python

Number of participants


Price per participant

Testimonials (7)

Upcoming Courses

Related Categories