- •Introduction
- •Background
- •Related work
- •Problem
- •Method
- •Chosen methods
- •Alternative methods
- •Implementation
- •Result
- •Quasi-experiment
- •Experiment
- •Survey (quantitative)
- •Survey (qualitative)
- •Discussion
- •Conclusion
- •Threats to validity
- •Ethics
- •Comparison to related work
- •Future work
- •Appendix
- •Appendix
- •Appendix
- •Appendix
3 | Result
This chapter contains the outcome of the two experiments as well as the survey. Key findings of the study, such as: descriptive statistics; inferential statistics; and relevant graphs with included Standard Deviation (SD), Standard Error (SE) and Confidence Interval (CI) are presented.
3.1Quasi-experiment
The following figures represent the results of the execution time for encoding and decoding of the two different libraries.
Figure 3.1: Execution time to encode (raw data)
18
Figure 3.2: Execution time to decode (raw data)
The remaining figures was created without filtering, such as removing outliers, of the raw data (Figure 3.1 & 3.2).
19
Figure 3.3: Execution time means to encode, with SD
Figure 3.4: Execution time means to decode, with SD
20
When comparing the SDs to the mean values of each bar (bitrate), for both encode and decode (Figure 3.3 & 3.4), aptX has a larger span than LC3. For example, when decoding, aptX (384) has a mean of 1.501 with an SD of 0.071 whilst LC3 (256) has a mean of 0.615 with an SD of 0.011. Hence, the SD of aptX (384) is approximately 4.7% of its mean whilst the SD of LC3 (256) is around 1.8%.
Figure 3.5: Execution time means to encode, with SE
21
Figure 3.6: Execution time means to decode, with SE
Since the SE for encoding and decoding (Figure 3.5 & 3.6) are quite small the sample mean should be a fair approximation of the true population. As an example, when decoding, aptX (384) has a SE of 0.007 and LC3 (256) has a SE of 0.001.
22
Figure 3.7: Execution time means to encode, with a CI of 95%
Figure 3.8: Execution time means to decode, with a CI of 95%
23
