### Table 3: Summary of Forecasting Methods

1997

"... In PAGE 15: ... One measurement was taken every thirty seconds over a twenty-four hour period beginning at 6:00 PM on Thursday, September 19, 1996. To investigate the relationship between model complexity and accuracy, we compare the predictive performance of a model tted using Semi-Nonparametric Time Series Analysis (SNP) developed by Gallant and Tauchen [10, 9] to the current set of NWS forecasting methods (shown in Table3 ). A complete description of each NWS technique is given in [16].... In PAGE 15: ... The model tting phase considers stationary autoregression in various forms (both with Gaussian and non-Gaussian noise terms), autoregressive, conditionally heteroscedastic (ARCH) models (with and without Gaussian noise), and general non-linear processes with heterogeneous innovations. Table 4 compares the performance of forecasts generated using SNP, with the techniques shown in Table3 that are implemented by the NWS.... ..."

Cited by 84

### Table 5: Point Forecast Evaluation: Multiple Forecast Encom- passing Test

"... In PAGE 15: ... The MSMH model seems to be second-best at the shorter forecast horizons whereas the SETAR model takes the second place at the longer forecast horizons. As seen from Table5 , the null hypothe- sis that the AR model forecast encompasses all other models simultaneously is rarely rejected. However, the same is true for the MSMH and MSIAH models and for the SETAR model at longer forecast horizons.... ..."

### Table 2: shows the percentage of archived forecasts, no forecasts and the percentage of

"... In PAGE 26: ... Observations and model data havetobeavailable. Table2 shows the percentage of archived forecasts for each month. The table also shows the pertentage of archived forecasts where road data hasn apos;t been available.... ..."

### Table 3 Comparison of forecasting capabilities

"... In PAGE 9: ...ay-ahead forecasts (see www.caiso.com). The performance of the model is summarized in Table3 (column 3). Looking at the mean square error (MSE) values for the whole test period we can observe that the CAISO forecast outperforms our model.... ..."

### Table 2: Forecast performance comparison.

"... In PAGE 20: ... We use the lter provided in Hamilton (1989) to compute smoothed conditional probabilities Pr[sT = 0jYT ; xT ]. Table2 shows the results of the forecast comparison. The rst two columns show the root mean squared forecast error [RMSE] and the mean absolute percentage forecast error [MAPE].... ..."

### Table 4 Forecasting performance measures

1999

"... In PAGE 9: ... Hence, we will use their versions in absolute terms. We provide in Table4 the... In PAGE 10: ... The indicator name and lags used are shown on top of each column. Table4 shows first the sample average absolute values of the slope being predicted over the forecasting horizon, 1998:1-1998:12. This is the reference with which the forecast statistics could be compared to evaluate forecast quality.... ..."

### Table 1a: Forecast Data

1998

"... In PAGE 3: ...eferences .............................................................................................................22 Appendix Table1 a.... In PAGE 4: ...LIST OF TABLES Table1 : Implied Standard Deviations and Vega apos;s .... In PAGE 4: ...able 6: Forecast Extremes ................................................................................19 Table1... In PAGE 17: ...1 For reliability reasons, prices for any option with a volume less than five were deleted. Table1 lists the remaining data (for informational purposes, the volumes are included in the table as well). On August 29 the closing futures price for Henry Hub natural gas was $2.... In PAGE 18: ... Table1 : Implied Standard Deviations and Vega apos;s NYMEX closing data from August 29, 1997 Contract price being forecasted: Oct. 97 Natural Gas Futures contract value: $2.... In PAGE 23: ... The monthly forecasts were then compared to the actual values of the gas prices on the dates being forecast. The predicted and actual prices for each forecast date are listed in Appendix Table1 a. Each forecast was derived from mid-month options prices.... ..."

### Table 1 Forecasts of period demands

2003

"... In PAGE 9: ... 5. A numerical example To illustrate the technique we shall use the demand forecasts presented in Table1 . The initial inventory level is taken as zero.... ..."

### TABLE 6 Distribution of the forecasting errors

"... In PAGE 14: ... TABLE 5 Parameter estimation (after diagnostic checks) and root mean square error for the two STARMA models. TABLE6 Distribution of the forecasting errors LIST OF FIGURES FIGURE 1 Loop detectors at the Athens road network. The ones used in this study are highlighted with different color and a label.... ..."

### Table 3 Forecast of #01f

"... In PAGE 14: ...Thus, wehave three competing forecasts, CM 1 ,N 1 ,P 1 for t + 1 and four, CM 2 ,N 2 ,NC 2 , and P 2 , for t +2. TABLE 3 ABOUT HERE We summarize the results in Table3 where we report both absolute and quadratic loss criteria #28MAPE and MSPE#29, and the Diebold and Mariano #28D-M#29 statistic S 1 , to test for the signi#0Ccance of the di#0Berence in MAPEs and MSPEs between our conditional model CM i , #28i=1,2#29 as a benchmark against eachof the alternative models. The sign of the estimated S 1 indicates whether the cor- responding forecast is better #28positive sign#29 or worse #28negative sign#29 than the benchmark in each subset.... ..."