| No. of records per page: 10 | 20 | 30 | 50 | 100 | Show all |
Record ID: 40 [ Page 13 of 16, No. 1 ]
High dimensional nonparametric discrete choice model
Authors: Maureen Dinna D. Giron
Abstract:
The functional form of a model can be a constraint in the correct prediction of discrete choices. The fl exibility of a nonparametric model can increase the likelihood of correct prediction. The likelihood of correct prediction of choices can further be increased if more predictors are included, but as the number of predictors approaches or exceeds the sample size, more serious complications can be generated than the improvement in prediction. With high dimensional predictors in discrete choice modeling, we propose a generalized additive model (GAM) where the predictors undergo dimension reduction prior to modeling. A nonparametric link function is proposed to mitigate the deterioration of model fi t as a consequence of dimension reduction. Using simulated data with the dependent variable having two or three categories, the method is comparable to the ordinary discrete choice model when the sample size is suffi ciently large relative to the number of predictors. However, when the number of predictors exceeds substantially the sample size, the method is capable of correctly predicting the choices even if the components included in the model account for only 20% of the total variation in all predictors.
Keywords: discrete choice model; generalized additive model; high dimensional data; nonparametric model
Download this article:
Year: 2013 Vol.: 62 No.: 1
Record ID: 39 [ Page 13 of 16, No. 2 ]
Esstimation under purposive sampling with auxiliary variable
Authors: John Erwin Banez
Abstract:
A PPS purposive selection and estimation was studied. A purposive sampling proposed by Guarte and Barrios (2006) is used. Instead of SRS, PPS with auxiliary variable was used in the selection. Results were compared to SRS estimates and SRS purposive. Standard error and coefficient of variation of estimates were basis for comparison. It was shown that PPS purposive has comparable results with SRS purposive and both have better performance compared to SRS.
Keywords: purposive sampling; bootstrap; auxiliary variable
Download this article:
Year: 2013 Vol.: 62 No.: 1
Record ID: 38 [ Page 13 of 16, No. 3 ]
Sparse principal component regression
Authors: Joseph Ryan G. Lansangan
Abstract:
Modeling of complex systems is usually confronted with high dimensional independent variables. Econometric models are usually built using time series data that often exhibit nonstationarity due to the impact of some policies and other economic forces. Both cases are usually affected by the multicollinearity problem resulting to unstable least squares estimates of the linear regression coefficients. Principal component regression can provide solution, but in cases where the regressors are nonstationary or the dimension exceeds the sample size, principal components may yield simple averaging of the regressors and the resulting model is difficult to interpret due to biased estimates of the regression coefficients. A sparsity constraint is added to the least squares criterion to induce the sparsity needed for the components to reflect the relative importance of each regressor in a sparse principal component regression (SPCR) model. Simulated and real data are used to illustrate and assess performance of the method. SPCR in many cases leads to better estimation and prediction than conventional principal component regression (PCR). SPCR is able to recognize relative importance of indicators from the sparse components as predictors. SPCR can be used in modeling high dimensional data, as an intervention strategy in regression with nonstationary time series data, and when there is a general problem of multicollinearity.
Keywords: sparsity; high dimensionality; multicollinearity; nonstationarity; sparse principal components
Download this article:
Year: 2013 Vol.: 62 No.: 1
Record ID: 37 [ Page 13 of 16, No. 4 ]
Value-at-risk measures for the PSE index using hidden markov models
Authors: Joselito C. Magadia
Abstract:
VaR measures for the PSE index are estimated using an m-state normal-hidden Markov model. The estimation procedure will be done under an unconditional approach and a conditional approach. Backtesting will be done to assess how well the estimates performed.
Keywords: homogeneous, irreducible, aperiodic Markov Chain
Download this article:
Year: 2013 Vol.: 62 No.: 1
Record ID: 36 [ Page 13 of 16, No. 5 ]
Bootstrap estimation of the average household expenditure on personal care and effects of regional level
Authors: Jachelle Anne G. Dimapilis
Abstract:
This study explored the use of bootstrap in estimating households’ expenditure on personal care and effects at regional level. Bootstrap yields superior estimates compared to the estimates of simple random sampling without replacement (SRSWOR). Bootstrap estimates have lower variance than SRSWOR estimates. Bootstrap estimates have smaller percentage difference from the actual mean compared with SRSWOR estimates.
Keywords: survey sampling; predictive estimation; bootstrap estimation
Download this article:
Year: 2013 Vol.: 62 No.: 1
Record ID: 35 [ Page 13 of 16, No. 6 ]
Nonparametric transfer function model with localized temporal effect
Authors: John Carlo P. Daquis
Abstract:
A semiparametric transfer function model is proposed and estimated using the backfitting algorithm. Simulation studies indicated that the procedure provides robust estimates for the transfer function especially for short time series data. This provides a viable alternative to the parametric transfer function model that requires large number of time points to estimate a number of parameters of the model. Furthermore, in the presence of seasonality or structural change, the procedure generally yields more robust estimates of the transfer function model than the maximum likelihood estimates of the parameters of the parametric model.
Keywords: transfer function model; semiparametric model; backfitting; mixed models
Download this article:
Year: 2013 Vol.: 62 No.: 1
Record ID: 34 [ Page 13 of 16, No. 7 ]
Sampling from a Skewed Population: The Sampling Design of the 2011 Survey of Enterprises in the Philippines
Authors: Erniel B. Barrios
Abstract:
Keywords:
Download this article:
Year: 2012 Vol.: 61 No.: 2
Record ID: 33 [ Page 13 of 16, No. 8 ]
Robust Methods in Time Series Models with Volatility
Authors: Wendell Q. Campano
Abstract:
Volatility in time series data is often accounted into the model by postulating a conditionally heteroskedastic variance. In-sample prediction maybe satisfactory but the out-sample prediction is usually problematic. A test for presence of volatility through a nonparametric method is proposed. An estimation procedure for the stationary part of the model by integrating block bootstrap and AR-sieve into the forward search algorithm is also provided. Simulation studies indicated high power for the nonparametric procedure in detecting local volatilities. On the other hand, the estimation method generated robust estimates of the parameters of the time series model in the presence of temporary volatility.
Keywords: block bootstrap; AR-sieve; forward search algorithm; nonparametric test; volatility
Download this article:
Year: 2012 Vol.: 61 No.: 2
Record ID: 32 [ Page 13 of 16, No. 9 ]
Poisson Spatial Autoregression Modelling of Poverty Count Data in the Philippines
Authors: John Erwin S. Banez
Abstract:
Count data with skewed distribution and possible spatial autoregression (SAR) often causes difficulty in modelling. Violations on the assumptions in ordinary least squares (OLS) may occur. While Poisson regression can offer some remedy in modelling count data, it still does not take into account the spatial dependencies of the data. This paper uses general linear estimation via backfitting algorithm in Poisson-SAR of poverty count in the Philippines for 2000. The model is assessed based on comparison from other models and the actual poverty count (MAPE and poverty map). MAPE was lowest in Poisson-SAR compared to other models.
Keywords: spatial autoregression; backfitting algorithm; poisson regression
Download this article:
Year: 2012 Vol.: 61 No.: 2
Record ID: 31 [ Page 13 of 16, No. 10 ]
Nonparametric Bootstrap Estimation of the Population Ratio Using Ranked Set Sampling
Authors: Kevin Carl P. Santos; Charisse Mae I. Castillo; Reyna Belle d.S. de Jesus; Nina B. Telan; Crystal Angela P. Vidal
Abstract:
Ranked Set Sampling (RSS) yields unbiased and more reliable estimators of the population mean and proportion while keeping low costs. Using nonparametric bootstrap estimation, the efficiency of the ratio estimates using RSS with Simple Random Sampling (SRS) are compared. A simulation study accounting for the sampling rate, population size, population variance and correlation with the concomitant variable was conducted to compare RSS and SRS in estimating ratios. When ranking was done on the numerator characteristic, RSS generally performs better than SRS in terms of their relative bias. Likewise, in terms of precision, RSS generally produces better estimates when ranking was done on the numerator characteristic. On homogeneous populations, contrary to what was expected, RSS performed better over SRS. On heterogeneous populations, on the other hand, the two sampling designs are generally comparable
Keywords: population ratio; ranked set sampling; simple random sampling; nonparametric bootstrap estimation
Download this article:
Year: 2012 Vol.: 61 No.: 2