Data Characteristics

[Nimtaa]

Hi, thank you for sharing the dataset. I am working on driving event detection using inertial sensors. I have logged some data through my IOS device (SensorLog application). I wanted to compare my generated signals with your dataset. I was wondering if you could let me know regarding some of the characteristics of your data such as scale and maximum range.

[jair-jr]

Hi! Thank you for your interest in the driver behavior dataset. I suggest you calculate descriptive statistics (min, max, quantiles) directly from the dataset. For example, you could generate boxplots for each axis of each sensor for both datasets. You could then compare the boxplots generated for my dataset with the ones generated for you dataset. Please, notice that I translated the data points from the device coordinate system to earth coordinate system. I used a rotation matrix as described here. So, please take that into consideration when comparing the datasets.

[Nimtaa]

Thank you for your suggestions. However, I did not find out how you translated the coordination system to the standard one (Earth). I have rotated your data through monitoring the data trends for each event.

[jair-jr]

Please, notice that the sensor data on the dataset is already rotated. I did not store the raw (unrotated) sensor data. The data rotation takes place in the android app in real time while sensor data is being collected. The rotation code is in class EarthAxesConverterFunction which uses a rotation matrix returned by SensorManager.getRotationMatrix method.

[HHorimoto]

Thank you for sharing the dataset. I have a question about the "Translate to Earth coordinate system" related to this issue in your paper. What you say above is that this preprocessing is done on all the sensor data in the data set? So if I want to follow your method in your paper, I just need to start from "Generate attribute vectors"? Thank you in advance!