Skip to content

Latest commit

 

History

History
100 lines (73 loc) · 4.27 KB

CodeBook.md

File metadata and controls

100 lines (73 loc) · 4.27 KB

Original Data


This project uses the following files from the UCI HAR Dataset:

  • activity_labels.txt - activity labels
  • features_info.txt - variable names
  • subject_test.txt - subject IDs from test group
  • subject_train.txt - subject IDs from train group
  • Y_test.txt - activity IDs from test group
  • Y_train.txt - activity IDs from train group
  • X_test.txt - measurements from test group
  • X_train.txt - measurements from train group

This original data provided 10299 values across 561 measured variables and 3 fixed variables (subject, activity, group).

Transformations performed to clean up the data


Subject, activity and measurements were combined in a single dataframe, adding a new 'group' column to identify whether measurements were from the training or test group. Activity IDs were replaced with lables from the 'activity_labels' file, and variables columns were named using the 'features_info' file.

This dataframe was subset to only include variables that measured the mean or standard deviation.

Variable names were modified for ease of understanding. Variable names are composed of several features, see table below.

The data was then grouped by subject, activity and test/train group, and the average calculated for each variable.

The data was restructured from wide to long with new columns for 'variable' and 'average' to form a simple tidy dataset "tidy_simple.txt".

Variables


Fixed variables

Variable Description
Subject ID of the subject that performed the activity
Activity activity performed, see Activity labels table
Group train or test group

Measured variable

Variable feature Description
Domain time or frequency
Component body or gravity
Instrument accelerometer o gyroscope
Jerk derived from acceleration or velocity
Magnitude magnitude of each 3 dimensional signal
Statistic mean or standard deviation
Axis XYZ direction components of the signal vector

Values

Data is presented as the average of each measurement for each variable grouped by subject, activity and group.

Activity labels

ID Label
1 WALKING
2 WALKING_UPSTAIRS
3 WALKING_DOWNSTAIRS
4 SITTING
5 STANDING
6 LAYING

Data structure

The tidy dataset has 11880 rows and 5 columns.

> str(tidy_simple)
'data.frame':	11880 obs. of  5 variables:
 $ subject : int  2 4 9 10 12 13 18 20 24 2 ...
 $ activity: Factor w/ 6 levels "LAYING","SITTING",..: 1 1 1 1 1 1 1 1 1 2 ...
 $ group   : Factor w/ 2 levels "test","train": 1 1 1 1 1 1 1 1 1 1 ...
 $ variable: Factor w/ 66 levels "Time.Body.Acc.mean.X",..: 1 1 1 1 1 1 1 1 1 1 ...
 $ average : num  0.281 0.264 0.259 0.28 0.26 ...

> head(tidy_simple)
   subject activity group             variable   average
 1       2   LAYING  test Time.Body.Acc.mean.X 0.2813734
 2       4   LAYING  test Time.Body.Acc.mean.X 0.2635592
 3       9   LAYING  test Time.Body.Acc.mean.X 0.2591955
 4      10   LAYING  test Time.Body.Acc.mean.X 0.2802306
 5      12   LAYING  test Time.Body.Acc.mean.X 0.2601134
 6      13   LAYING  test Time.Body.Acc.mean.X 0.2767164

 > tail(tidy_simple)
       subject         activity group                  variable    average
 11875      25 WALKING_UPSTAIRS train Freq.Body.GyroJerkMag.std -0.8410610
 11876      26 WALKING_UPSTAIRS train Freq.Body.GyroJerkMag.std -0.6698463
 11877      27 WALKING_UPSTAIRS train Freq.Body.GyroJerkMag.std -0.7517143
 11878      28 WALKING_UPSTAIRS train Freq.Body.GyroJerkMag.std -0.7048528
 11879      29 WALKING_UPSTAIRS train Freq.Body.GyroJerkMag.std -0.7564642
 11880      30 WALKING_UPSTAIRS train Freq.Body.GyroJerkMag.std -0.7913494