More Predictive Analytics · Microsoft Excel (Brianne Kwasny's Library) by Carlberg, Conrad -- Read -- Imperial Library of Trantor

Index

About This eBook Title Page Copyright Page Contents at a Glance Contents About the Author Dedication We Want to Hear from You! Reader Services Introduction 1. Smoothing and Its Alternatives

Regression in Forecasting

Curvilinear Relationships Piecewise Regression Polynomial Regression TREND() and LINEST() Using the Chart’s Trendline

Regression or Smoothing?

Comparing Smoothing with Regression with Forecasting Errors Additional Rationales for Smoothing Pushing the Forecast Horizon

2. Diagnosing Trend and Seasonality

Inside the Autocorrelation

The Magnitude and the Direction of a Correlation The Definitional Formula for Pearson’s Correlation Autocorrelation and Averages Looking at Individual Cases

Testing for Trend in the Observations

Understanding the Meaning of ACFs Interpreting the ACFs Using the ACF Add-in Testing All the ACFs at Once

Testing for Seasonality in a Time Series

Removing Trend Independent Residuals Analyzing the Regression with LINEST() Applying the Models Comparison Approach Time Series Diagnostics and the Analysis of Covariance

3. Working with Trended Time Series

Smoothing: The Basic Idea

The Smoothing Equation’s Error Correction Form Using Excel’s Solver to Choose Alpha The Equation’s Smoothing Form

About the Smoothing Approaches to Forecasting

A Time Series’ Pattern Additive and Multiplicative Models Initializing Values Initializing Versus Validation Terminology Problems More Smoothing Terminology

Dealing with Trend: Holt’s Linear Exponential Smoothing

Dealing with Trend by Differencing Forecasting the Differences

Using Holt’s Linear Exponential Smoothing

Forecasting a Level Forecasting a Trend Forecasting the Next Level and the Next Trend Extending the Forecast Optimizing the Constants Going Beyond the Next-Step-Ahead Forecast Validating the Results

4. Initializing Forecasts

Setting Initial Values Getting an Initial Forecast with Backcasting Getting an Initial Forecast with Optimization

Initial Forecast in Simple Exponential Smoothing Initial Forecasts with Holt’s Method Understanding the Relationship Between Holt’s Method and Linear Regression About Multistart and Solver

Initializing with Regression

Arranging for Curvilinear Regression Deriving the Regression Equation Confirming the Regression Equation

Deciding to Use Polynomial Regression

5. Working with Seasonal Time Series

Simple Seasonal Averages

Identifying a Seasonal Pattern Calculating Seasonal Indexes Forecasting from Simple Seasonal Averages: No Trend Simple Seasonal Averages with Trend Evaluating Simple Averages

Moving Averages and Centered Moving Averages

Using Moving Averages Instead of Simple Averages Understanding Specific Seasonals Aligning the Moving Averages Centered Moving Averages with Even Numbers of Seasons Detrending the Series with Moving Averages

Linear Regression with Coded Vectors

About Effect Coding and Regression Effect Coding with Seasons Setting Up the Coded Vectors Dummy Coding

Simple Seasonal Exponential Smoothing

About the Level Smoothing Formulas About the Season Smoothing Formulas Dealing with the End of the Time Series

Holt-Winters Models

Initializing the Forecasts Starting the Forecasts Earlier

6. Names, Addresses, and Formulas

We Interrupt This Program...

Implicit Intersections The Relative Range Name

Establishing Names for Forecasting Formulas

Establishing the Fixed Range Names Establishing the Mixed Range Names

Using the Names in Forecasting Formulas

Interpreting the Formula for Series Level Interpreting the Formula for the Seasonal Effect Interpreting the Forecast Formula

Deriving the Error Correction Formulas

The Level Equation The Trend Equation The Season Equation Demonstrating the Formulas on the Worksheet

Deriving the Formulas on the Worksheet

Deriving Error Correction of Level and Season Effects Deriving Error Correction of Level and Trend

Named Ranges for Holt-Winters Models

Estimating the Series Level in Holt-Winters Models Estimating the Current Trend Estimating the Current Seasonal Effect

Error Correction Formulas for Holt-Winters Models

7. Multiplicative and Damped Trend Models

About Multiplicative Models

Additive Models and Multiplicative Models Why Use an Additive Model?

Multiplicative and Additive Models Compared

Comparing the Level Estimates Comparing the Estimates of Seasonal Effects Producing the Next Forecast Using the Error Correction Formulas

Additive and Multiplicative Seasonal Models in Holt-Winters Analysis

Holt-Winters Additive Models Revisited Multiplicative Holt-Winters Models Changing the Initial Seasonal Estimates Modifying the Level and Seasonal Smoothing Formulas Modifying the Trend Smoothing Formula The Multiplicative Model with Error Correction Formulas

Damped Trend Forecasts

Formulas for the Damped Trend Implementing the Damped Trend on the Worksheet

Index Code Snippets

← Prev
Back
Next →

← Prev
Back
Next →