Count Distributions For Autoregressive Conditional Duration Model

P105- Count Distributions for Autoregressive Conditional Duration Models with Application to
Financial Data S.H. Ong , University of Malaya Atanu Biswas, Indian Statistical Institute S. Peiris, The
University of Sydney In this presentation the count distribution arising from the autoregressive
conditional duration (ACD) model of Engle and Russell (1998) is derived when the duration in the
ACD model has a generalized Weibull distribution (Mudholkar et al, 1996). The computation of its
probabilities is examined. Applications of the derived count distribution to analysis of price changes
in simulated and real stock transaction data and the number of polio cases are then considered.
Using a simulation study we show that the method of moments provides better estimates of the
parameters of interest. 42 P141- An Optimal Covariate Adjusted Response Adaptive procedure for
Normal Treatment Responses Rahul Bhattacharya, West Bengal State University Uttam
Bandopadhyay, University of Calcutta An optimal covariate adjusted adaptive allocation procedure is
developed for two treatment continuous response clinical trial. The proposed procedure maximizes
the benefited number of subjects and maintains a specified level of power for testing equality of
treatment effects. Assuming normality of the response variables several exact and asymptotic
properties of the proposed design are studied and compared with the covariate ignored counterpart.
P168- Development of the Utility Function of an Airline Travel: A Logarithmic Goal Programming
Approach Goutam Dutta, Priyanko Ghosh Indian Institute of Management Ahmedabad An airline
passenger has a number of choices before his/her travel decision. Competing airlines are also
interested regarding the preference set of the traveler. Passenger choice modeling is an essential
component of any Revenue Management System. In this paper we attempt to develop a linear utility
model for an airline travel using a logarithmic goal programming method. The utility score of the
competing airlines can help the passengers to choose the best feasible option among several
alternatives. The rank also can provide a system of mechanism that compares the competing airlines
in a common framework. We also derive the choice probability of the airlines by a multinomial logit
choice model. This facilitates the airlines to make a relative comparison among them and to estimate
the market share. 43 P085- A Rough Set Approach to develop an Efficient ldiversity Algorithm based
on Clustering B. K. Tripathy, VIT University
G. K. Panda, MITS K. Kumaran, VIT University Most of the organizations
publish micro data for a variety of purposes including demographic and public health research. To
protect the anonymity of the entities, data holders often remove or encrypt explicit identifiers. But,
released information often contains quasi identifiers, which leak valuable information. Samarati and
Sweeney introduced the concept of k‐anonymity to handle this problem and several algorithms have
been introduced by different authors in recent times. Lin et al put forth a new clustering‐based
method known as OKA for k‐anonymization. But, k‐anonymity can create groups that leak
information due to homogeneity attack. This problem is tackled by the notion of l‐ diversity
introduced by Machanavajjhala et al. Recently, the OKA algorithm is improved by Tripathy et al by
making some modifications in the adjustment stage and introducing distinct l‐diversity into it. But, in
most of the modern databases impreciseness has become a common characteristic, which is not
handled by any of the above algorithms. The primary purpose of this paper is to use MMeR, an
algorithm introduced by Tripathy et al, in developing a suitable anonymisation algorithm which is
applicable to any database having precise or imprecise heterogeneous data and satisfies both k‐
anonymity as well as l‐diversity properties. 44 P036- Efficient algorithm to predict rare events in a
temporal domain KoteswaraRao Kolli, Alekhya Viswanadhuni, Shilpa Kadam Prithvi Information
Solutions Limited In general, efficiency and accuracy of algorithms are the major concerns while
working with time series data. Here our focus is to identify potential patterns that precede rare
events and be able to predict occurrences of rare events within the network efficiently and
accurately. A literature review shows that defining temporal window size is considered as user
defined. This might be a setback as there is a possibility of losing some information which might
actually be of value in detecting hidden information. Here, we propose new algorithms to overcome
the problem stated by defining Dynamic window concept. Existing algorithms in the literature use
Apriori algorithm for predicting the frequent patterns in the temporal data, which is computationally
time consuming. To increase the efficiency of these algorithms we employ Prefix tree based search
method. We applied these proposed techniques on event log data of IT infrastructure as well as used
synthetic data sets to explore additional patterns, which were not found by the existing algorithms.
However, to evaluate if the additional patterns obtained might be of value to the network operators
or in general created noise, we have confirmed the same from domain expert and observed that
some of the patterns were extremely useful. Our statistics shows that the proposed algorithms have
out‐performed the existing techniques in both efficiency and accuracy aspects

Count Distributions For Autoregressive Conditional Duration Model

Uploaded by

Copyright:

Available Formats

Count Distributions For Autoregressive Conditional Duration Model

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Count Distributions For Autoregressive Conditional Duration Model

Uploaded by

Copyright:

Available Formats

P105- Count Distributions for Autoregressive Conditional Duration Models with Application to

You might also like