-
Notifications
You must be signed in to change notification settings - Fork 0
allenhsin/MobileActivity
Folders and files
Name | Name | Last commit message | Last commit date | |
---|---|---|---|---|
Repository files navigation
請按照以下步驟處理Data 1. sh combineRaw.sh 把多份的./raw/Accel* 資料combine為 ./rawData 2. java ParseRawData <rawData> <parsedRawDATA> 把rawData轉換成適合MATLAB吃的格式 default請打 java ParseRawData rawData parsedRawData 3a. java extractFeature <PARSEDRawData> <featureoutput> (not used) 吃兩個argument 第一個是rawData 第二個是output feature的檔名 p.s. compile java code first: javac extractFeature.java 把rawData 變成尚未normalize 的 feature 目前產生15 + 2 個feature label 給的方式是用voting 該timewindow中最多的activity做為該label 3b. extractFeature.m 新的feature extraction file 目前有7個feature 4. normalizeFeature.m 把各feature都normalize成最大值1 5. normFeaturePCA.m 把normFeatures變成principle components 6. plotFeature.m 把feature兩兩畫在2D space上,另外也把最重要的兩個PC畫在2D space上 7. python separator.py ## 要修正裡面兩個變數的值 1. NUM_OF_FEATURES 2.INPUT_FILE_NAME 將feature檔 依照groundtruth分成五個類 class0 -> 都是stationary的feature 以此類推 8. python sepTrainTest.py 將各個class的normalized feature分成train和test兩份 9. python combine.py 將各個class的train和test data合併的單一的train和test file 10. boostingClassification.m 使用adaboost classifier來作classification 11. plotClassifiedResult.m 把boosting的結果畫出來 12. boostingClassificationPCAPreTest.m 使用adaboost做Principle component的classification 並且決定dimension reduction的feature個數 13. plotClassifiedResultPCAPreTest.m 換成PCA的plot result 14. boostingClassificationPCA.m dimension reduction後用adaboost做classification 15. plotClassifiedResultPCA.m 換成PCA的result 16. java calMeanWithAllData (not used) 算各個cluster的mean
About
ML Project
Resources
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published