Entity

Time filter

Source Type

Bangalore, India

Chandra Katari A.,GENPACT India Pvt. Ltd | Umar Shaik N.,CSC Inc. | Rao Pasupuleti V.,P.A. College
International Journal of Applied Environmental Sciences | Year: 2013

In this paper we discussed a brief empirical application of ARIMA in various time intervals of US inflation data and provided recommendations on the time limit for ARIMA. Traditional time series models such as ARIMA models have been proven to be inadequate for modeling long or short range dependence. In this context we introduced MCARIMA methodology and its applications of fine-tuning of ARIMA residuals using Markov Chain to improve prediction accuracy and to decide the time interval for better forecasting. © Research India Publications. Source


Chandra Katari A.,GENPACT India Pvt. Ltd | Umar Shaik N.,CSC Inc. | Rao Pasupuleti V.,P.A. College
International Journal of Applied Environmental Sciences | Year: 2013

In this study, we use different techniques to predict the probability of customer payment behavior for accounts receivables (in the form of invoices) in finance business. Our goal in this paper is to develop a statistical method that yields predictive distributions for delinquency occurrence of the customer. For these purpose two different approaches, logistic regression and discriminant analysis are used and we also applied both methods in the case of unified and sitespecific scenarios. Results are presented and discussed to choose the best model. Using simple logistic regression and discriminant analysis, we illustrate the importance of comparing models with different number of parameters. Goodness of fit of the logistic regression model will be examined by using likelihood ratio test & wald's test. For discriminant function Eigen values (λ), canonical correlation eta (η), wilks' lambda (Λ) and chi-square test (χ2) are used for goodness of fit test. © Research India Publications. Source

Discover hidden collaborations