Buying the best firewall equiptment, installing the best anti-virus software, adding the greatest intrusion detection system, but you will still have something missing in your security plan that you may have overlooked. In this particular piece of analysis we will discuss what more you can do to protect your organization’s network.
(Kyung Hee University, 2007)
1. Independent T-Test
Any Statistical test is a t-test, in which we have t-distribution if the null hypothesis is true. This test is used when the sample size is so small that assumptions don’t work. Descriptive statistics for networks are calculated using T-test, but not used when measuring these statistics with standard errors.
Among the most frequently used t tests are:
· A test in which two normally distributed populations is equal. There are further two versions of T-test:
o Independent T-test: e.g. individuals randomly put into two groups
o Paired T-test: e.g. the same people measured before and after interference
2. One-Way ANOVA
When we come across one independent and one continues value, we use the one-way Analysis of Variance. The independent variable can consist of any number of Levels. An experimenter believes that studying in groups is better than studying alone, so he picked up kids at random and place them into different groups. Group one is assigned with a particular book, with only one child in it (controlled), group two had two children group three had three. After a calculated period of time, children were told to give an exam of which they studied.
Firstly notice, each experimental condition there is a different subject design. Second, by noticing instead of two levels of the independent variable, we now have three. This test is used when there are two levels of one independent variable. A multi-way ANOVA would be required, in situation like, there is a categorical independent variable and a continuous dependent variable and there are more than two levels of the independent variable and/or there is more than one independent variable.
Post Hoc Tests: Series of t-tests, which are kind of a Post Hoc Tests, were run to answer a series of Turkey’s, pair comparisons questions. The post-hoc tests are more severe than the regular t-tests however, is just because the more test you perform the more chance that you will find a significant difference just by luck. Your post hoc tests with statistical programs often presented in a table may look something like this:
Multiple comparison tests: to find out whether our calculated means are equal from the other means of not, the one-way analysis of variance is useful. This test is useful when comparing the mean, pair wise.
3. Mann-Whitney U Test
Sometimes we don’t see the normal distribution, or the samples taken are so small that one cannot tell if they are part of a normal distribution or not. Using the t-test to tell if there is a significant difference between samples is not appropriate here so we need something to provide us with the information needed. This test is useful in sitations like these where samples are very small. It can also be used when the variable being recorded is measured using an arbitrary scale which cannot be measured accurately e.g. a color scale measured by eye or a behavioral trait such as anger.
The paired Student’s t-test for the case of two related samples, we use The Wilcoxon signed-rank test. Like the t-test, it involves comparisons of the differences between measurements, so it requires the data tp be measured at an interval level of measurement. This test is applicable when the estimations of the t-test can’t be verified.
4. Kruskal-Wallis H Test
This test is derived from Wilcoxcon test, and it is used to test that the unpaired samples have been originated from the population. Codes are broken using the MedCalc factor to break up one variable into different sample subgroups. Most of the times “analysis of variance by ranks”, is the term given to the Kruskal-Wallis test to described as an (SSFA) Signal Strength Fourier Analysis, technique uses a (STFT) Short-term Fourier Transform to detect spoofing behavior.
5. Chi-square test
Qing Li and Prof Wade Trappe tried detecting a fault by using statistic performed on a sliding window of 250 signal strength, hence coming out with a Chi-square test. SSFA correctly picks out fault that the Chi-Square Test does not find these attacks at all, which are not visible to the human eye.
(Kyung Hee University, 2007)
6. Regression Analysis
The objective of this particular subject is to develop and test a model of online protection behavior, mainly regarding to the use of virus protection. Using a survey carried out in a collage of 273 students who use the Internet, a test of the hypotheses is conducted using multiple regression analysis. The result tell us that the perceived self-efficacy in using virus protection measures, perceived response efficacy of virus protective measures, positive outcome the protection measure taken for a virus threat, what do they think as a virus treat, their behavior before a virus hits them. The conclusion is that those who are in charge of the information security should not only take care of the increasing individuals’ awareness of the likelihood of virus attacks, but also conduct interventions aimed at increasing self-efficacy and response efficacy beliefs.
The main goal of regression analysis is to tell us the values of parameters for a function that cause the function to best fit a set of data observations that you are provided with. In linear regression, the function is a linear (straight-line) equation.
Regression equation is used to estimate, the difference between the observed value, and the value predicted is calculated by the residual analysis. The residual answers are calculated from the available data, and are treated as estimates. They are often used by statisticians to verify the assumptions.
1. Holden, Greg (2003). Guide to Firewalls and Network Security: Intrusion Detection and VPN,. Course Technology.
2. Kyung Hee University, K.H (2007). A respond system against DDos. Retrieved June 4, 2008, Web site: www.tc.apii.net/APIIworkshop2004/HP/pypark.pdf