Stochastic Data Model Research Articles

Combining Algorithmic and Stochastic Data Models: Application to Commercial Real Estate Prices

Combining Algorithmic and Stochastic Data Models: Application to Commercial Real Estate Prices

Combining Algorithmic and Stochastic Data Models

Combining Algorithmic and Stochastic Data Models

High Level Estimation of Power Consumption in Point-to-Point Interconnect Architectures

As technology shrinks, the importance of the communication architecture in the overall system performance and power consumption increases dramatically. In this work, a framework is developed to estimate the consumption in point-to-point interconnect structures at high levels of abstraction. To model the effect of cross coupled capacitances, the spatial correlationbetween adjacent wire lines is considered together with the transition activity, and both are efficiently estimated using word-level statistics. Based on a set of increasing complexity stochastic data models, an analytical estimation procedure is proposed and validated with both synthetic and real data sets. Extensive bit level simulations have been carried out to show the accuracy of the proposed models.

Utility-Privacy Tradeoffs in Databases: An Information-Theoretic Approach

Ensuring the usefulness of electronic data sources while providing necessary privacy guarantees is an important unsolved problem. This problem drives the need for an analytical framework that can quantify the privacy of personally identifiable information while still providing a quantifiable benefit (utility) to multiple legitimate information consumers. This paper presents an information-theoretic framework that promises an analytical model guaranteeing tight bounds of how much utility is possible for a given level of privacy and vice-versa. Specific contributions include: 1) stochastic data models for both categorical and numerical data; 2) utility-privacy tradeoff regions and the encoding (sanization) schemes achieving them for both classes and their practical relevance; and 3) modeling of prior knowledge at the user and/or data source and optimal encoding schemes for both cases.

The impact of ownership unbundling on cost efficiency

Several countries around the world have introduced reforms to the electric power sector. One important element of these reforms is the introduction of an unbundling process, i.e., the separation of the competitive activities of supply and production from the monopole activity of transmission and distribution of electricity. There are several forms of unbundling: functional, legal and ownership. New Zealand, for instance, adopted an ownership unbundling in 1998. As discussed in the literature, ownership unbundling produces benefits and costs. One of the benefits may be an improvement in the level of the productive efficiency of the companies due to the use of the inputs in just one activity and a greater level of transparency for the regulator. This paper analyzes the cost efficiency of 28 electricity distribution companies in New Zealand for the period between 1996 and 2011. Using a stochastic frontier panel data model, a total cost function and a variable cost function are estimated in order to evaluate the impact of ownership unbundling on the level of cost efficiency. The results indicate that ownership separation of electricity generation and retail operations from the distribution network has a positive effect on the cost efficiency of distribution companies in New Zealand. The estimated effect of ownership separation suggests a positive average one-off shift in the level of cost efficiency by 0.242 in the short-run and 0.144 in the long-run.

An inverse solution of the lifetime-oriented design problem

This paper presents a new solution of the lifetime-oriented design problem. This solution is based on a point-to-point allocation between the space of the design parameters and the space of structural responses. Each point in the space of the design parameters defines a feasible or non-feasible design, and all feasible designs guarantee compliance with a predetermined lifetime. From the set of feasible designs, one or more designs may be selected with the aid of technical or economic criteria. The presented solution permits the consideration of non-statistical data uncertainty, thereby leading to an uncertain lifetime. Because of the unavoidable information deficit, for example incomplete data in practical problems, the application of non-statistical data uncertainty is more realistic than the application of stochastic data models. The selection of feasible design variants is based on methods of explorative data analysis.

Modeling of Data Reduction in Wireless Sensor Networks

In this paper, we present a stochastic model for data in a Wireless Sensor Network (WSN) using random field theory. The model captures the space-time behavior of the underlying phenomenon being observed by the network. We present results regarding the size and spatial distribution of the regions of the network that sense statistically extreme values of the underlying phenomenon using the theory of extreme excursion regions. These results compliment many existing works in the literature that describe algorithms to reduce the data load, but lack an analytical approach to evaluate the size and spatial distribution of this load. We show that if only the statistically extreme data is transmitted in the network, then the data load can be significantly reduced. Finally, a simple performance model of a WSN is developed based on a collection of asynchronous M/M/1 servers that work in parallel. We derive several performance measures from this performance model. The presented results will be useful in the design of large scale sensor networks.

Statistical Modeling: The Two Cultures (with comments and a rejoinder by the author)

There are two cultures in the use of statistical modeling to reach conclusions from data. One assumes that the data are generated by a given stochastic data model. The other uses algorithmic models and treats the data mechanism as unknown. The statistical community has been committed to the almost exclusive use of data models. This commitment has led to irrelevant theory, questionable conclusions, and has kept statisticians from working on a large range of interesting current problems. Algorithmic modeling, both in theory and practice, has developed rapidly in fields outside statistics. It can be used both on large complex data sets and as a more accurate and informative alternative to data modeling on smaller data sets. If our goal as a field is to use data to solve problems, then we need to move away from exclusive dependence on data models and adopt a more diverse set of tools.

The kriging update model and recursive space-time function estimation

We present a method for efficiently fitting a time series of spatial functions to observed data. The method is closely related to kriging, which is an interpolation method based on a stochastic data model. While kriging is effective and versatile for estimating individual functions from observed data, it must be extended to incorporate temporal correlation. In this paper, we introduce temporal correlation to kriging in the form of a stochastic state equation representation-the kriging update model. This permits a recursive solution that is akin to Kalman filtering to estimate time series of functions that avoids growing data problems associated with other space-time extensions of kriging. The state equation representation incorporates the principle assumption of universal kriging: that the mean is deterministic but unknown. We derive the estimate using best linear unbiased estimation and state the result in a concise algorithm for general use in arbitrary spatial dimensions. To demonstrate the algorithm, we apply it to two sets of test functions and provide an example application in estimating heart motion from medical images.