UUM ETD | Universiti Utara Malaysian Electronic Theses and Dissertation
FAQs | Feedback | Search Tips | Sitemap

Classification of Capital Expenditures and Revenue Expenditures Using Neural Network Model

Abolgasim, Adnan Ali (2008) Classification of Capital Expenditures and Revenue Expenditures Using Neural Network Model. Masters thesis, Universiti Utara Malaysia.

[img] PDF
Restricted to Registered users only

Download (783kB)

Download (171kB) | Preview


Financial statements are the basic means of transferring a complete picture about any company's performance which is the basement for any business decision making process. Hence, it is very important to ensure their healthy condition and to diagnose and cure any problem they might suffer from. One of the common problems is the over/under stating of the profit/loss and assets figures in the financial statements due to the misclassification between the capital and revenues expenditures. Moreover, it is practically difficult in some cases to draw a line between the two types. This study aims to integrate an Artificial Intelligence technique such as Neural Networks in order to develop a model that can be trained to recognize hidden patterns of the borderline between the two expenditures types. Thus it can successfully help in the classification between the capital and revenue expenditures. Twelve criterions were identified in order to classify between the two expenditures types and a Multi Layer Perceptron (MLP) was incorporated in the constructed neural network model using Neural Connection 2.0. The dataset was collected based on various cases of capital and revenue expenditures. The classification accuracy of the model was 97.51% for training and 94.20% for testing. Analysis has shown a significant correlation between identified criterions (input variables) with model target. Strong correlation between target and criterion LASMFY (0.532), which indicates that any expenditure lasts for more than a fiscal year will be more probable to be classified into a capital expenditure. Same conclusion goes to criterion RESORGN with a strong correlation of (-0.539), which indicates any expenditure that restored an existing asset to its original operating capacity will have more probability to be classified into a capital expenditure as well. Also, criterion RESALE proves its strong influence, since its correlation was (-0.874) this implies more probability of classification into revenue expenditure if any expenditure was spent for intent for resale. Medium correlation shown by criterion REGULR (-0.251) indicates a moderate probability of classification into revenue expenditure if expenditure was spent in a regular basis. Criterions with a weak correlation represent less probability of classification into either of the two expenditures types (capital or revenue expenditure), which implies that each of theses criterions is heavily depending on contribution of the rest of criterions in order to be able to classify. These conclusions were found to be in line with definitions of capital and revenues expenditure drawn by accounting authors such as (Al-Daif, 1981) and (Fess & Warren, 1987).

Item Type: Thesis (Masters)
Uncontrolled Keywords: Artificial Intelligence, Neural Networks, Financial Statements, Capital Expenditures, Revenues Expenditures
Subjects: Q Science > QA Mathematics > QA76 Computer software
Divisions: College of Arts and Sciences (CAS)
Depositing User: Mr Husni Ismail
Date Deposited: 07 Dec 2009 02:00
Last Modified: 24 Jul 2013 12:09
URI: http://etd.uum.edu.my/id/eprint/891

Actions (login required)

View Item View Item