Permalink
Cannot retrieve contributors at this time
| # | |
| # feat-bug-corr-7dig.R, 20 Mar 20 | |
| # | |
| # Various analysis of http://www.7digital.com feature data | |
| # | |
| # Example from: | |
| # Evidence-based Software Engineering: based on the publicly available data | |
| # Derek M. Jones | |
| # | |
| # TAG | |
| source("ESEUR_config.r") | |
| plot_layout(2, 1) | |
| par(mai=c(1, 0.8, 0.0, 0.1)) | |
| source(paste0(ESEUR_dir, "projects/agile-work/feat-common-7dig.R")) | |
| # For each day return the total number of work-days for each | |
| # feature that has reached a Done 'state' | |
| feature_work_increment=function(done_list) | |
| { | |
| done_num=as.integer(done_list$Done)-base_day | |
| dl=rep(0, end_day) | |
| sum_done=function(done_day) | |
| { | |
| dl[done_day] <<- dl[done_day]+sum(done_list$Cycle.Time[done_num == done_day]) | |
| } | |
| dummy=sapply(1:end_day, function(x) sum_done(x)) | |
| return(dl) | |
| } | |
| hol_days=as.vector(get_ph_days())-base_day | |
| hol_days=subset(hol_days, (hol_days > 0) & !is.na(hol_days)) | |
| # Plot cross-correlation between bugs and non-bugs | |
| #all_bug_prio=sum_starts(subset(p, Type == "Production Bug")$Done) | |
| all_bug_prio=sum_starts(subset(p, grepl(".*Bug$", p$Type))$Prioritised) | |
| all_bug_prio=all_bug_prio[-weekends] | |
| all_bug_prio=all_bug_prio[-hol_days] | |
| all_bug_prio=rollmean(all_bug_prio, 3) | |
| all_features=feature_work_increment(p) | |
| all_features=all_features[-weekends] | |
| all_features=all_features[-hol_days] | |
| all_features=rollmean(all_features, 3) | |
| non_bug_done=feature_work_increment(subset(p, !grepl(".*Bug$", p$Type))) | |
| #non_bug_done=sum_starts(subset(p, !grepl(".*Bug$", p$Type))$Done) | |
| #non_bug_done=sum_starts(subset(p, Type == "MMF")$Done) | |
| non_bug_done=non_bug_done[-weekends] | |
| non_bug_done=non_bug_done[-hol_days] | |
| non_bug_done=rollmean(non_bug_done, 3) | |
| # Many bugs reported at start will be caused by feature implementations | |
| # that were Done before the start of recording. Ignore the | |
| # first 100 workdays so we are probably only checking for a correlation | |
| # where one might exist | |
| ccf(non_bug_done[-(1:100)], all_bug_prio[-(1:100)], lag.max=150, | |
| xlab="Work days (nonbug features)", ylab="Cross correlation\n", | |
| main="") | |
| ccf(all_features[-(1:100)], all_bug_prio[-(1:100)], lag.max=150, | |
| xlab="Work days (all features)", ylab="Cross correlation\n", | |
| main="") | |