Difference between revisions of "ISSS608 2017-18 T3 Assign Wang Runyu Data Visualization"
Jump to navigation
Jump to search
Line 33: | Line 33: | ||
=====Correlation Plot===== | =====Correlation Plot===== | ||
+ | remove features based on corr plot observations | ||
+ | |||
[[File:Corr plot q2 visual method.png|none|600px]] | [[File:Corr plot q2 visual method.png|none|600px]] | ||
+ | Corrplot, select variables, we not only consider to check the variables nearby, but all the variables | ||
+ | For example, group of 'specSlape_sd', 'harmonics_median', 'peakFreq_median', I choose 'peakFreq_median'. | ||
=====Trellis Plot for Variable Density===== | =====Trellis Plot for Variable Density===== |
Revision as of 11:36, 7 July 2018
|
|
|
|
Contents
Q1
ph
Q2
Visual Analytical Approach
As the text format outcome from seewave package(analyzeFolder function) contains 69 features, it is not practical for us to analyze all the features visually. In the first step, I use findCorrelation() function to eliminate highly correlated features. I set the pair-wise absolute correlation cut off to 0.6.
Correlation Plot
remove features based on corr plot observations
Corrplot, select variables, we not only consider to check the variables nearby, but all the variables For example, group of 'specSlape_sd', 'harmonics_median', 'peakFreq_median', I choose 'peakFreq_median'.