Clean Data

Assumsion

• the files (activity_labels.txt, features.txt) have to be in the same workspace with the run_analysis.R
• two folder test and train need to be in same workspace with the script file
• library reshape2 and plyr installed

How run_analysis script work

• Collect the information of activity labels and features (activity_labels.txt, features.txt)

• Get data from test folder:

  • collect data from test/subject_test.txt using read.table, set column name as subjects
  • collect data from test/X_test.txt using read.table, set column names as featureNames [features.txt]
    • take subset of the features data by checking features name contains mean or std words
    • write down the origin feature names [meanAndStdFeatures.txt] to help users to map back to the origin data because set column names in R have filter out all special characters.
  • collect data from test/y_test.txt using read.table, set column names as activities
    • convert the activity codes into activity label
      • convert activity codes to numeric
      • Using the numeric as index to get correct label
  • combind all data by column

• Get data from train folder:

  • collect data from train/subject_train.txt using read.table, set column name as subjects
  • collect data from train/X_train.txt using read.table, set column names as featureNames [features.txt]
    • take subset of the features data by checking features name contains mean or std words
    • write down the origin feature names [meanAndStdFeatures.txt] to help users to map back to the origin data because set column names in R have filter out all special characters.
  • collect data from train/y_train.txt using read.table, set column names as activities
    • convert the activity codes into activity label
      • convert activity codes to numeric
      • Using the numeric as index to get correct label
  • combind all data by column

• Combind test and train data by row

• Write tidyData

• Create data set with the average of each variable for each activity and each subject.

  • reshape data by using melt on activities and subjects
  • for each activity and subject, apply mean on the value column
  • write down the result in averageMovementByActivityAndSubject.txt