The estimate of US Natural Gas well head price (the
dependent variable) have been attempted using values of
independent variables viz. US Crude Oil first purchase price
and Gold average price by applying time series and non
parametric regression. Nonparametric regression
methodology called Alternating Conditional Estimation
(ACE) algorithm and Time series model using
Autoregressive integrated moving average (ARIMA) has
been used to derive functional relationship between the
dependent (response) variable and multiple independent
(predictor variables). All time series models except
GARCH provide well-defined and comparable values of
forecast against the observed values. The closeness in
estimated and observed values corroborates that the
estimated model is able to provide adequate fit to the data.
However, the ARIMA (1, 2, 1) model using standard errors
based on Hessian seems to outperform other time series
models. The transform obtained by ACE has an optimal
regression coefficient of 0.95 and the satisfactory fit is
observed between measured and that derived from
transform.