About Course
On the course, we will sequentially go through all the stages of working with data: from the types of tasks and their formulation to working with machine learning models to minimize the predictive error. Additionally, we will consider the fundamental principles of building machine learning models, basic metrics and the simplest models – linear and logistic regression.
Also, consider the basic linear models and all the practical aspects of using linear regression to predict ASHRAE energy numbers.
We will analyze classification metrics and models, and then we will work out applied approaches to data classification using Prudential insurance scoring machine learning models and ensembles.
Let’s analyze the segmentation and classification of cloud images using convolutional, pyramidal, residual and fully connected neural networks.
Course Content
Part 1: The Machine Learning Process
-
Machine learning tasks
-
Machine learning tasks
-
Machine learning model and process
-
What is ETL
-
Machine learning process
-
What is EDA
-
Data preparation
-
Data preparation
-
Splitting the sample
-
Hyperparameter Optimization
-
Latin square (hypercube)
-
Hyperparameter Optimization via Parzen Trees
-
Undertraining and overtraining
-
Bias, Scatter, and Data Error
-
Model training
-
Using HDF
Metrics and Models
Part 2. Workshop: Prediction of energy consumption of buildings
Workshop: Memory optimization and data enrichment
Linear regression models
Workshop: Competitive regression models
Workshop: Linear Regression Ensemble
Part 3. Metrics and classification models
Workshop: Insurance scoring problem
Simple Classification Models
Workshop: Logistic regression and support vector machines
Ensemble models
Workshop: Decision trees and ensembles of bagging and boosting
Advanced Ensembles
Workshop: Staking ensemble and final solution
Part 4. Artificial neural networks
Workshop: Cloud Shape Recognition
Neural network training
Workshop: Convolutional neural networks
Convolutional neural network architectures
Workshop: Neural Network Architectures
Workshop: neural networks for segmentation
Student Ratings & Reviews
