National Library of Energy BETA

Sample records for multinomial logit nmnl

  1. A new method for multinomial inference using Dempster-Shafer theory

    Office of Scientific and Technical Information (OSTI)

    (Journal Article) | SciTech Connect A new method for multinomial inference using Dempster-Shafer theory Citation Details In-Document Search Title: A new method for multinomial inference using Dempster-Shafer theory A new method for multinomial inference is proposed by representing the cell probabilities as unordered segments on the unit interval and following Dempster-Shafer (DS) theory. The resulting DS posterior is then strengthened to improve symmetry and learning properties with the

  2. A Plug-in Hybrid Consumer Choice Model with Detailed Market Segmentation

    SciTech Connect (OSTI)

    Lin, Zhenhong; Greene, David L


    This paper describes a consumer choice model for projecting U.S. demand for plug-in hybrid electric vehicles (PHEV) in competition among 13 light-duty vehicle technologies over the period 2005-2050. New car buyers are disaggregated by region, residential area, attitude toward technology risk, vehicle usage intensity, home parking and work recharging. The nested multinomial logit (NMNL) model of vehicle choice incorporates daily vehicle usage distributions, refueling and recharging availability, technology learning by doing, and diversity of choice among makes and models. Illustrative results are presented for a Base Case, calibrated to the Annual Energy Outlook (AEO) 2009 Reference Updated Case, and an optimistic technology scenario reflecting achievement of U.S. Department of Energy s (DOE s) FreedomCAR goals. PHEV market success is highly dependent on the degree of technological progress assumed. PHEV sales reach one million in 2037 in the Base Case but in 2020 in the FreedomCARGoals Case. In the FreedomCARGoals Case, PHEV cumulative sales reach 1.5 million by 2015. Together with efficiency improvements in other technologies, petroleum use in 2050 is reduced by about 45% from the 2005 level. After technological progress, PHEV s market success appears to be most sensitive to recharging availability, consumers attitudes toward novel echnologies, and vehicle usage intensity. Successful market penetration of PHEVs helps bring down battery costs for electric vehicles (EVs), resulting in a significant EV market share after 2040.

  3. A new method for multinomial inference using Dempster-Shafer...

    Office of Scientific and Technical Information (OSTI)

    Furthemore, posterior inference on relative probabilities amongst certain cells depends only on data for the cells in question. Finally, the model is quite flexible with regard to ...

  4. A new method for multinomial inference using Dempster-Shafer...

    Office of Scientific and Technical Information (OSTI)

    Number: AC52-06NA25396 Resource Type: Journal Article Resource Relation: Journal Name: Annals of Statistics Research Org: Los Alamos National Laboratory (LANL) Sponsoring Org: DOE...

  5. Predicting primary crystalline phase and liquidus temperature above or below 1050{degrees}C as functions of glass composition

    SciTech Connect (OSTI)

    Redgate, P.E.; Piepel, G.F.


    This report presents the results of applying statistical empirical modeling techniques to primary crystalline phase at the liquidus temperature (T{sub L}) and (ii) whether liquidus temperature is above or below 1050{degree}C (1OO{degree}C below a melting temperature of 1150{degree}C). Data used in modeling primary crystalline phase and liquidus temperate are from the Composition Variability Study (CVS) of Hanford waste glass compositions and properties. The majority of the 123 CVS glasses are categorized into one of 13 primary crystalline phases (at the liquidus temperature). They are also classified as to having T{sub L} Above or Below 1050{degree}C. Two common statistical methods used to model such categorical data are the multinomial logit and classification tree models. The classification tree models provided an overall better modeling approach than did the multinomial logit models. The performance of models in this report should be compared to the performance of the revised ``Development of Models and Software for Liquidus Temperature of Glasses of HWVP Products`` models from Ecole Polytechnique. If the Ecole Polytechnique models perform better than the models discussed in this report, no additional effort on these models would be needed. However, if the converse is true, it may be worthwhile to invest additional effort on statistical empirical modeling methods.

  6. ADOPT: A Historically Validated Light Duty Vehicle Consumer Choice Model

    SciTech Connect (OSTI)

    Brooker, A.; Gonder, J.; Lopp, S.; Ward, J.


    The Automotive Deployment Option Projection Tool (ADOPT) is a light-duty vehicle consumer choice and stock model supported by the U.S. Department of Energy’s Vehicle Technologies Office. It estimates technology improvement impacts on U.S. light-duty vehicles sales, petroleum use, and greenhouse gas emissions. ADOPT uses techniques from the multinomial logit method and the mixed logit method estimate sales. Specifically, it estimates sales based on the weighted value of key attributes including vehicle price, fuel cost, acceleration, range and usable volume. The average importance of several attributes changes nonlinearly across its range and changes with income. For several attributes, a distribution of importance around the average value is used to represent consumer heterogeneity. The majority of existing vehicle makes, models, and trims are included to fully represent the market. The Corporate Average Fuel Economy regulations are enforced. The sales feed into the ADOPT stock model. It captures key aspects for summing petroleum use and greenhouse gas emissions This includes capturing the change in vehicle miles traveled by vehicle age, the creation of new model options based on the success of existing vehicles, new vehicle option introduction rate limits, and survival rates by vehicle age. ADOPT has been extensively validated with historical sales data. It matches in key dimensions including sales by fuel economy, acceleration, price, vehicle size class, and powertrain across multiple years. A graphical user interface provides easy and efficient use. It manages the inputs, simulation, and results.

  7. Comparative Evaluation of Two Methods to Estimate Natural Gas Production in Texas

    Reports and Publications (EIA)


    This report describes an evaluation conducted by the Energy Information Administration (EIA) in August 2003 of two methods that estimate natural gas production in Texas. The first method (parametric method) was used by EIA from February through August 2003 and the second method (multinomial method) replaced it starting in September 2003, based on the results of this evaluation.

  8. Modeling patterns in count data using loglinear and related models

    SciTech Connect (OSTI)

    Atwood, C.L.


    This report explains the use of loglinear and logit models, for analyzing Poisson and binomial counts in the presence of explanatory variables. The explanatory variables may be unordered categorical variables or numerical variables, or both. The report shows how to construct models to fit data, and how to test whether a model is too simple or too complex. The appropriateness of the methods with small data sets is discussed. Several example analyses, using the SAS computer package, illustrate the methods.


    SciTech Connect (OSTI)

    Liu, Changzheng; Greene, David L


    How demand for E85 might evolve in the future in response to changing economics and policies is an important subject to include in the National Energy Modeling System (NEMS). This report summarizes a study to develop an E85 choice model for NEMS. Using the most recent data from the states of Minnesota, North Dakota, and Iowa, this study estimates a logit model that represents E85 choice as a function of prices of E10 and E85, as well as fuel availability of E85 relative to gasoline. Using more recent data than previous studies allows a better estimation of non-fleet demand and indicates that the price elasticity of E85 choice appears to be higher than previously estimated. Based on the results of the econometric analysis, a model for projecting E85 demand at the regional level is specified. In testing, the model produced plausible predictions of US E85 demand to 2040.

  10. An Evaluation of Parametric and Nonparametric Models of Fish Population Response.

    SciTech Connect (OSTI)

    Haas, Timothy C.; Peterson, James T.; Lee, Danny C.


    Predicting the distribution or status of animal populations at large scales often requires the use of broad-scale information describing landforms, climate, vegetation, etc. These data, however, often consist of mixtures of continuous and categorical covariates and nonmultiplicative interactions among covariates, complicating statistical analyses. Using data from the interior Columbia River Basin, USA, we compared four methods for predicting the distribution of seven salmonid taxa using landscape information. Subwatersheds (mean size, 7800 ha) were characterized using a set of 12 covariates describing physiography, vegetation, and current land-use. The techniques included generalized logit modeling, classification trees, a nearest neighbor technique, and a modular neural network. We evaluated model performance using out-of-sample prediction accuracy via leave-one-out cross-validation and introduce a computer-intensive Monte Carlo hypothesis testing approach for examining the statistical significance of landscape covariates with the non-parametric methods. We found the modular neural network and the nearest-neighbor techniques to be the most accurate, but were difficult to summarize in ways that provided ecological insight. The modular neural network also required the most extensive computer resources for model fitting and hypothesis testing. The generalized logit models were readily interpretable, but were the least accurate, possibly due to nonlinear relationships and nonmultiplicative interactions among covariates. Substantial overlap among the statistically significant (P<0.05) covariates for each method suggested that each is capable of detecting similar relationships between responses and covariates. Consequently, we believe that employing one or more methods may provide greater biological insight without sacrificing prediction accuracy.

  11. Electric and hybrid electric vehicles: A technology assessment based on a two-stage Delphi study

    SciTech Connect (OSTI)

    Vyas, A.D.; Ng, H.K.; Santini, D.J.; Anderson, J.L.


    To address the uncertainty regarding future costs and operating attributes of electric and hybrid electric vehicles, a two stage, worldwide Delphi study was conducted. Expert opinions on vehicle attributes, current state of the technology, possible advancements, costs, and market penetration potential were sought for the years 2000, 2010, and 2020. Opinions related to such critical components as batteries, electric drive systems, and hybrid vehicle engines, as well as their respective technical and economic viabilities, were also obtained. This report contains descriptions of the survey methodology, analytical approach, and results of the analysis of survey data, together with a summary of other factors that will influence the degree of market success of electric and hybrid electric vehicle technologies. Responses by industry participants, the largest fraction among all the participating groups, are compared with the overall responses. An evaluation of changes between the two Delphi stages is also summarized. An analysis of battery replacement costs for various types is summarized, and variable operating costs for electric and hybrid vehicles are compared with those of conventional vehicles. A market penetration analysis is summarized, in which projected market shares from the survey are compared with predictions of shares on the basis of two market share projection models that use the cost and physical attributes provided by the survey. Finally, projections of market shares beyond the year 2020 are developed by use of constrained logit models of market shares, statistically fitted to the survey data.

  12. Assessment Of Carbon Leakage In Multiple Carbon-Sink Projects: ACase Study In Jambi Province, Indonesia

    SciTech Connect (OSTI)

    Boer, Rizaldi; Wasrin, Upik R.; Hendri, Perdinan; Dasanto,Bambang D.; Makundi, Willy; Hero, Julius; Ridwan, M.; Masripatin, Nur


    Rehabilitation of degraded forest land throughimplementation of carbon sink projects can increase terrestrial carbonstock. However, carbon emissions outside the project boundary, which iscommonly referred to as leakage, may reduce or negate the sequestrationbenefits. This study assessed leakage from carbon sink projects thatcould potentially be implemented in the study area comprised of elevensub-districts in the Batanghari District, Jambi Province, Sumatra,Indonesia. The study estimates the probability of a given land use/coverbeing converted into other uses/cover, by applying a logit model. Thepredictor variables were: proximity to the center of the land use area,distance to transportation channel (road or river), area of agriculturalland, unemployment (number of job seekers), job opportunities, populationdensity and income. Leakage was estimated by analyzing with and withoutcarbon sink projects scenarios. Most of the predictors were estimated asbeing significant in their contribution to land use cover change. Theresults of the analysis show that leakage in the study area can be largeenough to more than offset the project's carbon sequestration benefitsduring the period 2002-2012. However, leakage results are very sensitiveto changes of carbon density of the land uses in the study area. Byreducing C-density of lowland and hill forest by about 10 percent for thebaseline scenario, the leakage becomes positive. Further data collectionand refinement is therefore required. Nevertheless, this study hasdemonstrated that regional analysis is a useful approach to assessleakage.

  13. Targeted chlorination for biofouling control in steam electric power plant condenser tubes

    SciTech Connect (OSTI)

    Tewari, R.N.


    The objective of this study was to develop an understanding of the relationships between biofouling, heat transfer, and to assess the technical feasibility of the targeted chlorination (TC) concept as an alternative to conventional chlorination in once-through seawater cooling systems. A shell and tube type heat exchanger was designed and constructed. The test facility monitored biofouling growth by measuring heat transfer resistance (thermal performance), pressure drop (hydraulic performance), and biomass deposit. Biocide effectiveness was determined by triplicates average residual biomass. Tests were run to evaluate 18 treatment scenarios. Preliminary tests were done to determine variabilities between shells and among tubes, effect of heating, very high dose chlorine burn out and regrowth behavior. LOGIT, a software package by the National Bureau of Standards (NBS) for sigmoidal fit, was used for the preliminary test data. The other three tests studied effect of chlorine dosage (low, high), contact time, and frequency. To minimize variability in results, triplicates for each chlorination scheme were used. The total amount of chlorine applied (flow rate Q x C x T x F) ranged from about 400 to 4,000 pounds per million gallons per day (an equivalent to a CTF range of 6 to 60 ppm.min/day). With respect to efficiency of biofouling control, it was found by regression analysis that among four variables C, T, (CT) and F, F was most significant and C, T, and (CT) were relatively equal in ranking.

  14. Normal Tissue Complication Probability Modeling of Radiation-Induced Hypothyroidism After Head-and-Neck Radiation Therapy

    SciTech Connect (OSTI)

    Bakhshandeh, Mohsen; Hashemi, Bijan; Mahdavi, Seied Rabi Mehdi; Nikoofar, Alireza; Vasheghani, Maryam; Kazemnejad, Anoshirvan


    Purpose: To determine the dose-response relationship of the thyroid for radiation-induced hypothyroidism in head-and-neck radiation therapy, according to 6 normal tissue complication probability models, and to find the best-fit parameters of the models. Methods and Materials: Sixty-five patients treated with primary or postoperative radiation therapy for various cancers in the head-and-neck region were prospectively evaluated. Patient serum samples (tri-iodothyronine, thyroxine, thyroid-stimulating hormone [TSH], free tri-iodothyronine, and free thyroxine) were measured before and at regular time intervals until 1 year after the completion of radiation therapy. Dose-volume histograms (DVHs) of the patients' thyroid gland were derived from their computed tomography (CT)-based treatment planning data. Hypothyroidism was defined as increased TSH (subclinical hypothyroidism) or increased TSH in combination with decreased free thyroxine and thyroxine (clinical hypothyroidism). Thyroid DVHs were converted to 2 Gy/fraction equivalent doses using the linear-quadratic formula with {alpha}/{beta} = 3 Gy. The evaluated models included the following: Lyman with the DVH reduced to the equivalent uniform dose (EUD), known as LEUD; Logit-EUD; mean dose; relative seriality; individual critical volume; and population critical volume models. The parameters of the models were obtained by fitting the patients' data using a maximum likelihood analysis method. The goodness of fit of the models was determined by the 2-sample Kolmogorov-Smirnov test. Ranking of the models was made according to Akaike's information criterion. Results: Twenty-nine patients (44.6%) experienced hypothyroidism. None of the models was rejected according to the evaluation of the goodness of fit. The mean dose model was ranked as the best model on the basis of its Akaike's information criterion value. The D{sub 50} estimated from the models was approximately 44 Gy. Conclusions: The implemented normal tissue complication probability models showed a parallel architecture for the thyroid. The mean dose model can be used as the best model to describe the dose-response relationship for hypothyroidism complication.