Machine Learning with R Course

Machine Learning with R

COURSE TRAINERS

Details

Machine Learning is a branch of computer science that uses algorithms to help artificial intelligent systems learn and adapt. That learning process requires data and a programming language to process it. R is a language built for data science which has many packages to for machine learning, ready for you to use.

In this course you will learn:

Key concepts and terms for machine learning
To remove noise in data via smoothing
To assess the effectiveness of your model with cross-validation
To integrate model development using the Caret Package
Useful algorithms for important tasks
To apply machine learning to real-world applications
How to work with large data sets

Delivery Methods

Leading Training is focusing on providing virtual training courses for the foreseeable future and will only consider in-person and classroom training on request, with a required minimum group size of six delegates. We remain committed to offering training that is fast, focused and effective.

Delivery Method	Duration	Price (excl. VAT)
Classroom	Days	ZAR 0.00	Get a Quote
Live Virtual Training	Days	ZAR 0.00	Get a Quote

Discounts Available

Brochure:

Information may change without notice.

Audience

Data scientists
Programmers
Business analysts
Engineers
Scientists

Pre-Requisites

Leading Training's Introduction to R Programming or equivalent knowledge

Course Outline / Curriculum

Introduction to machine learning

Notation
An example
Exercises
Evaluation metrics
- Training and test sets
- Overall accuracy
- The confusion matrix
- Sensitivity and specificity
- Balanced accuracy and F1 score
- Prevalence matters in practice
- ROC and precision-recall curves
- The loss function
Exercises
Conditional probabilities and expectations
- Conditional probabilities
- Conditional expectations
- Conditional expectation minimizes squared loss function
Exercises
Case study: is it a 2 or a 7?

Smoothing

Bin smoothing
Kernels
Local weighted regression (loess)
- Fitting parabolas
- Beware of default smoothing parameters
Connecting smoothing to machine learning
Exercises

Cross validation

Motivation with k-nearest neighbors
- Over-training
- Over-smoothing
- Picking the k in kNN
Mathematical description of cross validation
K-fold cross validation
Exercises
Bootstrap
Exercises

The caret package

The caret train functon
Cross validation
Example: fitting with loess

Examples of algorithms

Linear regression
- The predict function
Exercises
Logistic regression
- Generalized linear models
- Logistic regression with more than one predictor
Exercises
k-nearest neighbors
Exercises
Generative models
- Naive Bayes
- Controlling prevalence
- Quadratic discriminant analysis
- Linear discriminant analysis
- Connection to distance
Case study: more than three classes
Exercises
Classification and regression trees (CART)
- The curse of dimensionality
- CART motivation
- Regression trees
- Classification (decision) trees
Random forests
Exercises
Machine learning in practice
Preprocessing
k-nearest neighbor and random forest
Variable importance
Visual assessments
Ensembles
Exercises
Large datasets
- Matrix algebra
- Notation
- Converting a vector to a matrix
- Row and column summaries
- apply
- Filtering columns based on summaries
- Indexing with matrices
- Binarizing the data
- Vectorization for matrices
- Matrix algebra operations
Exercises
Distance
- Euclidean distance
- Distance in higher dimensions
- Euclidean distance example
- Predictor space
- Distance between predictors
Exercises
Dimension reduction
- Preserving distance
- Linear transformations (advanced)
- Orthogonal transformations (advanced)
- Principal component analysis
- Iris example
- MNIST example
Exercises
Recommendation systems
- Movielens data
- Recommendation systems as a machine learning challenge
- Loss function
- A first model
- Modeling movie effects
- User effects
Exercises
Regularization
- Motivation
- Penalized least squares
- Choosing the penalty terms
Exercises
Matrix factorization
- Factors analysis
- Connection to SVD and PCA
Exercises
34 Clustering
Hierarchical clustering
k-means
Heatmaps
Filtering features
Exercises

Schedule Dates and Booking

There are currently no scheduled dates.

Please note that this course needs a minimum of 6 delegates to schedule a course. You can choose to be added to the waiting list by clicking the button below, and we will contact you when we have enough delegates interested. Should we not get enough delegates, we will refund or credit your paid booking.

Add me to the waiting list

Should you need this course urgently, the following options are available:

Pay for 6 delegates (whether you have them or not) and we will schedule the course as soon as possible.
If you have fewer delegates and cannot pay for 6, we can negotiate a shortened course where some of the time will be spent in blended learning - watching videos and doing tutorials and exercises with some contact time with the trainer. We would want to discuss what your core needs are so that we cover those aspects. You need to have paid for 3 delegates at least.

Submit Enquiry