Correlation Analysis
Manasvi Thakur
February 19, 2016
The document illustrates correlation analysis for software components in R. The solution under consideration is Cisco Connected Roadways. This analysis takes into account 2 products namely: ISR 3900 & ISR 4400.
Methodology:
- SWDPMH data is downloaded for each product and saved as xlsx
- the R script reads reads the data using R package named xlsx.
- correlation analyses done for components that have an aggregate of 50 or more SR’s for the given time frame.
- the data used is for March 2014 - January 2016 time frame.
Overview of clean data for ISR 3900:
The columns represent the components and the rows represent the number of SR’s each month.
## isr-g2-sw ipipgw crypto-ace crypto-engine voice-sip ipsec-core
## X201403 12 3 7 4 3 9
## X201404 13 7 13 9 9 3
## X201405 10 7 6 9 3 0
## X201406 7 7 6 7 2 4
## X201407 8 5 5 7 8 8
## X201408 10 10 10 5 2 6
## cce-dp ribinfra c3900 voice-rtp voice-dsp ssl-vpn cme-sip eigrp
## X201403 6 0 4 3 3 2 0 5
## X201404 4 2 4 1 3 1 1 2
## X201405 2 1 3 0 4 5 0 3
## X201406 6 2 2 2 4 4 1 5
## X201407 8 3 3 1 7 3 1 2
## X201408 6 3 9 9 6 4 0 6
## ssh nat c2900
## X201403 0 5 0
## X201404 2 5 2
## X201405 3 3 0
## X201406 3 0 2
## X201407 3 3 4
## X201408 5 0 0Overview of clean data for ISR 4400:
## esg-chassismgr esg-ipsec esg-io-infra
## X201403 0 0 0
## X201404 0 0 1
## X201405 0 0 1
## X201406 0 2 1
## X201407 0 0 3
## X201408 0 0 0Now once we have processed data for both the products, we can run correlation analysis between them. We use R package - corrplot to plot our results.
The plot shows the correlations between components of two products.
Listed below are components that have correlation higher than 0.5.
## Var1 Var2 Freq
## 1 esg-chassismgr ribinfra 0.5610730
## 2 esg-io-infra cme-sip 0.6425864