added ggplot histograms of start times & episode durations

MTUS-W9-adult-episodes-data-processing.Rmd

@@ -68,6 +68,7 @@ reqLibsLocal <- c("foreign", # loading SPSS/STATA
        "data.table", # fast data manipulation
        "dplyr", # data manipulation
        "dtplyr", # data table & dplyr code
+       "ggplot2", # slick graphs
        "stargazer", # for pretty tables
        "knitr" # for kable
        )
@@ -662,14 +663,6 @@ Now the HH:MM version:
                                                 r_epStartM), 
                                          format = "%H:%M")
             ]
-  print("# Check distributions of episode starts (all surveys pooled)")
-  t <- mtusUKEpsDT[,
-                    .(
-                      N = length(epnum)
-                    ),
-                    by = r_epStartTime
-                      ]
-  plot(t)
 
 # create episode end
 mtusUKEpsDT$r_epEndTime <- mtusUKEpsDT$r_epStartTime + (mtusUKEpsDT$end*60)
@@ -681,6 +674,27 @@ Now the HH:MM version:
 summary(mtusUKEpsDT$r_epEndTime)
 ```
 
+```{r plotStartTimes, fig.cap = "Check distributions of episode starts by survey"}
+ggplot(mtusUKEpsDT) + 
+  geom_histogram(aes(x = r_epStartTime)) +
+  scale_x_datetime(date_labels = "%H:%M", date_breaks = "2 hours") +
+  facet_grid(survey ~ .) +
+    labs( 
+       caption = "MTUS UK samples"
+  )
+  
+```
+
+```{r plotDurations, fig.cap = "Check distributions of episode durations by survey"}
+mtusUKEpsDT <- mtusUKEpsDT[, epDurationMins := (r_epEndTime - r_epStartTime)/60]
+
+ggplot(mtusUKEpsDT) +
+  geom_histogram(aes(x = epDurationMins)) +
+  facet_grid(survey ~ .) +
+    labs(x = "Episode duration (minutes)", caption = "MTUS UK samples")
+  
+```
+
 ## Create half hours
 Create and check the distribution of half hours. This uses the epStartTime variable to make sure it is set for all cases, even those which still have no date: