Practical Time Series Forecasting with R

Contents Preface . . . 9 1 Approaching Forecasting . . . 15 1.1 Forecasting: Where? . . . 15 1.2 Basic Notation . . . 15 1.3 The Forecasting Process . . . 16 1.4 Goal Definition . . . 18 1.5 Problems . . . 23 2 Time Series Data . . . 25 2.1 Data Collection . . . 25 2.2 Time Series Components . . . 28 2.3 Visualizing Time Series . . . 30 2.4 Interactive Visualization . . . 35 2.5 Data Pre-Processing . . . 39 2.6 Problems . . . 42 3 Performance Evaluation . . . 45 3.1 Data Partitioning . . . 45 3.2 Naive Forecasts . . . 50 3.3 Measuring Predictive Accuracy . . . 51 3.4 Evaluating Forecast Uncertainty . . . 55 3.5 Advanced Data Partitioning: Roll-Forward Validation . . . 62 3.6 Example: Comparing Two Models . . . 65 3.7 Problems . . . 67 4 Forecasting Methods: Overview . . . 69 4.1 Model-Based vs. Data-Driven Methods . . . 69 4.2 Extrapolation Methods, Econometric Models, and External Information . . . 70 4.3 Manual vs. Automated Forecasting . . . 72 4.4 Combining Methods and Ensembles . . . 73 4.5 Problems . . . 77 5 Smoothing Methods . . . 79 5.1 Introduction . . . 79 5.2 Moving Average . . . 80 5.3 Differencing . . . 85 5.4 Simple Exponential Smoothing . . . 87 5.5 Advanced Exponential Smoothing . . . 90 5.6 Summary of Exponential Smoothing in R Using ets . . . 98 5.7 Extensions of Exponential Smoothing . . . 101 5.8 Problems . . . 107 6 Regression Models: Trend & Seasonality . . . 117 6.1 Model with Trend . . . 117 6.2 Model with Seasonality . . . 125 6.3 Model with Trend and Seasonality . . . 129 6.4 Creating Forecasts from the Chosen Model . . . 132 6.5 Problems . . . 133 7 Regression Models: Autocorrelation & External Info . . . 143 7.1 Autocorrelation . . . 143 7.2 Improving Forecasts by Capturing Autocorrelation: AR and ARIMA Models . . . 147 7.3 Evaluating Predictability . . . 153 7.4 Including External Information . . . 154 7.5 Problems . . . 170 8 Forecasting Binary Outcomes . . . 179 8.1 Forecasting Binary Outcomes . . . 179 8.2 Naive Forecasts and Performance Evaluation . . . 180 8.3 Logistic Regression . . . 181 8.4 Example: Rainfall in Melbourne, Australia . . . 183 8.5 Problems . . . 187 9 Neural Networks . . . 189 9.1 Neural Networks for Forecasting Time Series . . . 189 9.2 The Neural Network Model . . . 190 9.3 Pre-Processing . . . 194 9.4 User Input . . . 195 9.5 Forecasting with Neural Nets in R . . . 196 9.6 Example: Forecasting Amtrak Ridership . . . 198 9.7 Problems . . . 201 10 Communication and Maintenance . . . 203 10.1 Presenting Forecasts . . . 203 10.2 Monitoring Forecasts . . . 205 10.3 Written Reports . . . 206 10.4 Keeping Records of Forecasts . . . 207 10.5 Addressing Managerial "Forecast Adjustment" . . . 208 11 Cases . . . 211 11.1 Forecasting Public Transportation Demand . . . 211 11.2 Forecasting Tourism (2010 Competition, Part I) . . . 215 11.3 Forecasting Stock Price Movements (2010 INFORMS Competition) . . . 219 Data Resources, Competitions, and Coding Resources . . . 225 Bibliography . . . 227 Index . . . 231