Specialized Models: Time Series and Survival Analysis

IBM via Coursera

Coursera based on 133 ratings

Likes

Overview

This course introduces you to additional topics in Machine Learning that complement essential tasks, including forecasting and analyzing censored data. You will learn how to find analyze data with a time component and censored data that needs outcome inference. You will learn a few techniques for Time Series Analysis and Survival Analysis. The hands-on section of this course focuses on using best practices and verifying assumptions derived from Statistical Learning. By the end of this course you should be able...

Syllabus

Introduction to Time Series Analysis

This module introduces the concept of forecasting and why Time Series Analysis is best suited for forecasting, compared to other regression models you might already know. You will learn the main components of a Time Series and how to use decomposition models to make accurate time series models.

Stationarity and Time Series Smoothing

This module introduces you to the concepts of stationarity and Time Series smoothing. Having a Time Series that is stationary is easy to model. You will learn how to identify and solve non-stationarity. Smoothing is relevant to you as it will improve the accuracy of your models.

ARMA and ARIMA Models

This module introduces moving average models, which are the main pillar of Time Series analysis. You will first learn the theory behind Autoregressive Models and gain some practice coding ARMA models. Then you will extend your knowledge to use SARMA and SARIMA models as well.

Deep Learning and Survival Analysis Forecasts

This module introduces two additional tools for forecasting: Deep Learning and Survival Analysis. In addition to AI and Machine Learning applications, Deep Learning is also used for forecasting. Survival Analysis is a branch of Statistics first ideated to analyze hazard functions and the expected time for an event such as mechanical failure or death to happen. Survival Analysis is still used widely in the pharmaceutical industry and also in other business scenarios with limited data related to censoring, the lack of information on whether an event occurred or not for a certain observation.