2017 Fiscal Year Final Research Report
Distributon Characteristics of Financial Accounting Data and Exploring Analysis Methods
| Project/Area Number |
25380620
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Research Field |
Accounting
|
|---|
| Research Institution | Hosei University |
|---|
Principal Investigator |
|
|---|
| Project Period (FY) |
2013-04-01 – 2018-03-31
|
| Keywords | 冪分布 / 財務データ / フラクタル性 / テキストマイニング / ディープラーニング |
|---|
| Outline of Final Research Achievements |
Firstly, I confirmed the distribution characteristics of the figures such as total assets, net income, etc. of all listed companies in Japan, and also confirmed that these financial account values follow the power law. Time series characteristics of these values following power law are known to have fractal nature (self-similarity). So I have calculated fractal dimension of each financial value, but no satisfactory results are obtained due to sample size. Alternatively, I performed frequency analysis. I made histogram using part or the whole of data, then compared the shape of these histogram. Histogram using part of the data and histogram using the whole of data have very similar shape that shows both are subject to the power law. The observed distribution is power-distributed and indicated fractal nature strongly. In addition to these methods, I have also investigated the applicability of emerging methods such as Text Mining and Deep Learning.
|
| Free Research Field |
会計学
|
