Data cleaning - Githubissues

carissalow / rapids

Reproducible Analysis Pipeline for Data Streams

http://www.rapids.science/

GNU Affero General Public License v3.0

37 stars 20 forks source link

Data cleaning #166

Closed Meng6 closed 3 years ago

Meng6 commented 3 years ago

Updated the code for data cleaning:

Fill NA with 0 for the selected event phone features within each sensor's feature extraction script
Fill NA with 0 for the selected event phone features (do not consider plugin for this step, e.g. AR) in data cleaning script
Add drop highly correlated features in data cleaning script

Need to do before merging:

Update docs
Update testing results
Update example workflow

JulioV commented 3 years ago

Thanks @Meng6 Can you check the tests, please? They are falling for activity recognition

Meng6 commented 3 years ago

Thank you @JulioV. Weiyu is working on the testing results currently. Will let you know once it is done.

Meng6 commented 3 years ago

Hi @JulioV, the testing results, example workflow, and docs were updated. Could you please double check when you are free? Thanks!

Meng6 commented 3 years ago

Hi @JulioV, we just found that there might be an issue when time segment is very small. I split the [DATA_YIELDED_HOURS_RATIO_THRESHOLD] parameter into two parameters: DATA_YIELD_UNIT and [DATA_YIELD_RATIO_THRESHOLD]. I also updated the docs as you suggested. Could you double check the latest docs? If it is good, I can merge it back to the develop branch! Thank you!

JulioV commented 3 years ago

Hi @JulioV, we just found that there might be an issue when time segment is very small. I split the [DATA_YIELDED_HOURS_RATIO_THRESHOLD] parameter into two parameters: DATA_YIELD_UNIT and [DATA_YIELD_RATIO_THRESHOLD]. I also updated the docs as you suggested. Could you double check the latest docs? If it is good, I can merge it back to the develop branch! Thank you!

Let's cheange DATA_YIELD_UNIT to DATA_YIELD_FEATURE and possible values to RATIO_VALID_YIELDED_HOURS & RATIO_VALID_YIELDED_MUNUTES

Other than this and the comment above, it's ready