R and Statistics
Below are some sample work using R to do statistics and more.
Course Project / Individual Project
We introduced three methods in regression diagnosis: subset selection, scatterbox matrix and influence plot. Subset selection allows you to exclude extremely “bad” independent variables. Scatterbox matrix offers you a glance of whether there is any abnormal observation. Influence plot uses residual, hat and cook’s distance to exclude the extreme observations.
Bootstrapping in R
Bootstrapping is a common statistic approach. But how can we apply bootstrapping in mediation in R? Other than using Rmediation package, we can also code our own function.
Below is a function that applies bootstrapping to mediation analysis in using R code.
Matrix Computing Capacity with R and Microsoft R
This analysis serves to evaluate whether Microsoft R Open (using parallel processing) outperforms R (using a single processor at a time) in matrix computing. I use matrix inverse and matrix transpose as two tasks and use computation time as the criterion.
Analysis indicates that thought there is no significant difference in computing matrix transpose, Microsoft R Open has significant performance gains in computing matrix Inverse.