Experiment 1: From network to local user agent
Raw output from Microsoft Excel 2010 (Beta)
| Difference in RTP clock from previous sample | Inter-arrival times (in seconds) of RTP packets | |||
|---|---|---|---|---|
| Mean | 160 | Mean | 0.019999999 | |
| Standard Error | 0 | Standard Error | 9.28526E-08 | |
| Median | 160 | Median | 0.020004 | |
| Mode | 160 | Mode | 0.020005 | |
| Standard Deviation | 0 | Standard Deviation | 3.04446E-05 | |
| Sample Variance | 0 | Sample Variance | 9.26874E-10 | |
| Kurtosis | #DIV/0! | Kurtosis | 12.36652501 | |
| Skewness | #DIV/0! | Skewness | -2.054662184 | |
| Range | 0 | Range | 0.000374 | |
| Minimum | 160 | Minimum | 0.019815 | |
| Maximum | 160 | Maximum | 0.020189 | |
| Sum | 17200960 | Sum | 2150.11991 | |
| Count | 107506 | Count | 107506 | |
| Confidence Level(95.0%) | 0 | Confidence Level(95.0%) | 1.8199E-07 | |
Transcript
So, I can look at it [in terms of] the difference in the clocks from the previous sample, or I can look at it in terms of seconds. I can look at what the standard error is. I can look at the median, the mode, the standard deviation, the sample variance, kurtosis, skewness, etc. And skewness, tells us about the shape of the distribution. I can look at the range, the minimum, maximum, ... . I can sum up all the values. I can look at how many there were. And I can look at my confidence level at 95%, and I can see that I have this data with a hundred thousand samples, and I have a confidence level of about 1.8 x 10-7. So not surprisingly, with a hundred thousand samples - I'm fairly confident in this data - because I have such a large number of samples.