Recognize users of mobile devices from accelerometer data ( Accelerometer Biometric Competition on kaggle)
A course Project for 'CS 725: Foundations of Machine Learning'
#Description: You can find the description of problem statement at https://www.kaggle.com/c/accelerometer-biometric-competition
#How To Run the Code:
Assumption: You have train.csv and test.csv in the same folder as other project files
-
Run device_count.py
-
Run extractMeanVar.py
-
Run trimmingdata.py
-
Now you can execute any classifier code. Just look for the required files are there in the same folder.
#Our Approaches We devised following approaches for solving the problem:
-
Naive Bayes
-
Nearest Neighbors
-
Quadratic Discriminative Analysis ( Similar to LDA)
-
Support Vector Machines
#File related to each approach:
1] Naive Bayes
1. extractMeanVar.py
2. naive_bayes.py
3. naive_bayes_Random.py
2] Nearest Neighbors
1. trimmingdata.py
2. device_count.py
3. KNearestNeighbour.py
3] Quadratic Discriminative Analysis ( Similar to LDA)
1. qda.py
4] Support Vector Machines
1. svm_mean.py
2. svm_modified.py