Context The dataset contains preprocessed data from [CareerCon 2019](https://www.kaggle.com/c/career-con-2019) competition. The major difference from the original dataset is that it doesn't include quaternion columns but Euler angles differences instead. Also, the data was normalized to make it ready for Deep Learning models training. Content There are three NumPy arrays. 1. `feat.npy` with normalized training and testing data 2. `feat_fft.npy` with the same data but processed with `np.fft.rfft` call and then also normalized 3. `target.npy` with training labels and dummy zero labels for the test data concatenated into a single array Note that the first 3810 rows in every array belong to the training set, and the rest is testing data. Acknowledgements The data was originally prepared by [prith189](https://www.kaggle.com/prith189) in [this kernel](https://www.kaggle.com/prith189/starter-code-for-3rd-place-solution). Therefore, this dataset was created from the output and saved into NumPy arrays to simplify the creation of new kernels.