sit-stand

created intially by Kenneth Lai. ~May 2017

Please feel free to send any questions to: hykenneth.lai@berkeley.edu

File Descriptions

• adaptive.py*: Algorithms for step count
• dataReader.py: reads all data (sit, stand, walk) into dataFrames. The data files (.csv) are located in the data folder.
• realTimePred.py: functions for running real-time prediction
• realTimePredMain.py: run this file to test real-time prediction with the sensor.
• stepCountMain.py: takes in a dataFrame (walking) and does step count.
• utils.py: Lots of functions dealing with formatting data, creating test verification datasets, moving intervals, etc.
• verification.py: verification on several different motion protocols (for exmaple: switching for stand to sit). Also includes testHyperParam, the function that optimizes for threshold and interval.
• verificationContinuous.py: similar to above but uses a moving average dataset for testing
• verificationNew.py: functions for testing (w/o protocols). Produces confusion matrix and classification reports. Use testHyperParam to find the best threshold and interval to use in accuracyTest

To Run Verification

1. With designed protocols:
   1. See top of verification.py to see how to design protocols
   2. create Target with the function createPattern, located in verification.py
   3. run testHyperParam. datasets should be your recorded data for that specific protocol. patterns is the target that you created in the previous step. This step can be skipped.
   4. use testMotion, with your recorded dataset, target, interval, and threshold.
2. Without protocols (just verifying accuracy of raw data):
   1. Run accuracyTest located in verificationNew.py. This should take all data that is processed in dataReader.py.