Overview

Save Big on Coursera Plus. 7,000+ courses at $160 off. Limited Time Only!

Grab it

In this course you learn to develop and maintain a large-scale forecasting project using SAS Visual Forecasting tools. Emphasis is initially on selecting appropriate methods for data creation and variable transformations, model generation, and model selection. Then you learn how to improve overall baseline forecasting performance by modifying default processes in the system. This course is appropriate for analysts interested in augmenting their machine learning skills with analysis tools that are appropriate for assaying, modifying, modeling, forecasting, and managing data that consist of variables that are collected over time. The courses is primarily syntax based, so analysts taking this course need some familiarity with coding. Experience with an object-oriented language is helpful, as is familiarity with manipulating large tables.

Syllabus

Specialization Overview (Review)

In this module you get an overview of the courses in this specialization and what you can expect. Note: This same module appears in each course in this specialization.

Course Overview
Introduction to Large-Scale Forecasting

In this modules you'll get an overview of the functionality used in the course. We'll describe how objects and methods in the Automatic Time Series Modeling, or ATSM, package in SAS Visual Forecasting can be combined to solve the large-scale forecasting problem. We'll also describe how the configuration of objects and information flows change depending on what stage of the automatic forecasting process you are in.

Exploring and Processing Timestamped Data

In this module we'll use the TSMODEL procedure to perform time series accumulation and missing value interpretation. We'll use packages for PROC TSMODEL, which are blocks of code that can be inserted within the flow of your PROC TSMODEL code to perform specialized tasks for both data preparation and analysis. Then, we'll discuss time series hierarchies and how to use a BY statement in PROC TSMODEL to create a hierarchy.

Automatic Forecasting: Model Specification and Selection

In this module, we'll use the ATSM package in PROC TSMODEL to perform automatic forecasting, model selection, and specification. We'll walk through the process for declaring and using the many different ATSM objects and discuss how and where each object fits within the automatic forecasting process.

Creating Custom Models and Managing Model Lists

This module describes and illustrates functionality for creating your own custom models in the forecasting system. We'll provide step-by-step instructions for building a custom specification and then modifying the automatic model selection process to include your model as a candidate for all series in a given level of the data hierarchy.

Event Variables in the Forecasting System

In this module, we'll generate event variables three different ways. First, we'll use the ATSM package to create and implement predefined event variables. Second, we'll create event variables using the HPFEVENTS procedure. Third, we'll perform conditional BY-group processing for event variable creation. Next, we'll use and identify ARIMAX and ESM models, produce model selection lists, and select a champion model. Using the selected champion model and passing the predefined event variables to the TSMODEL procedure, we'll generate automatic forecasts and output model estimates and fit statistics.

Reconciling Statistical Forecasts

Reconciling statistical forecasts occurs after the automatic model generation, selection, and forecasting processes are done. In this module, we describe the reconciliation process and illustrate system tools and options for reconciling statistical forecasts we generated earlier in the course.

Setting Up the Forecasting System and Generating Best Forecasts

This module covers a variety of topics. First, we'll discuss system tools and best practices that have the potential to improve the precision of your system forecasts. These include best practices like honest assessment for champion model selection and system tools like outlier detection and combined model forecasts. Next, we'll describe options and best practices associated with rolling the system forward in time.