Store and compare android sensor(accelerometer) data sets to match the pattern

Question

I am developing android application which will learn and than evaluate aggressive driving behavior i.e. sharp turns, sudden brake etc...
As i am trying to make a learning algorithm. My approach now is to get sensor data when wrongly driven and make data sets of multiple sensors i.e. Accelerometer, Gyroscope and magnetometer.
Once multiple datasets(for each sensor) stored, the application will keep on comparing the datasets with the current values of the sensors. If pattern matched, same action(wrong) occurred.
Is there any api or solution to how can i compare and match the pattern of the sensor data, as currently i am using arrays and am comparing the multiple values which wont work, i know.
Or is there another solution to the problem which you think is better kindly lemme know.
It would be a great help as i am stuck here from weeks :(

Answer 1

This seem to be nlp problem where we need to find the patterns between the two data.You can make use of latent semantic analysis.There is one library gensim which is completely free to use. If the purpose is not solved by nlp approach then go for clustering. The clustering algorithm. Or if you have data only for one class (say when the diver is driving wrong) then use baysian approach.For one class it will work better.