In the "data" folder there is all the data needed to run the script run_analysis.R. The data is exactly the same as was provided in the .zip file.
##How run_analysis.R script works
- checks if there is dplyr package installed and installs it if it's not the case.
- loads all data the files into data frames
- merges test and train data, putting test data first
- names variables according to how they are defined in data/activity_labels.txt
- removes columns with duplicated names (as otherwise we'll have problems with dplyr later, and we don't need these columns anyway)
- chooses only columns with names that have "mean" and "std", case insensitive
- adds data about which subject corresponds to each row and which activity type (presented as string)
- makes subject and activity columns the first in the data frame
- saves the data to "dirty.txt"
- generates and saves (to "tidy.txt") a tidy dataset where for each pair (subject, activity) the mean of each variable (names are in data/activity_labels.txt) is calculated.
To explain the last point. Let's say we have following piced of data:
| subject | activity | var1 |
|---|---|---|
| 1 | ACTIVITY1 | VAR1_VALUE1 |
| 1 | ACTIVITY1 | VAR1_VALUE2 |
| 1 | ACTIVITY2 | VAR_VALUE3 |
Then in the file "dirty.txt" data will be saved as it is. In the file "tidy.txt" following data will be saved:
| subject | activity | var1 |
|---|---|---|
| 1 | ACTIVITY1 | mean(c(VAR1_VALUE1, VAR1_VALUE2)) |
| 1 | ACTIVITY2 | VAR_VALUE3 |
That is, since the pair subject="1" and activity="ACTIVITY1" was present twice, the mean of corresponding var1 values was taken
There is some more info in CodeBook.md