Gaussian Quantiles Datasets - DiRo2C

The two datasets are used to simulate two different black boxes that are supposed to predict different results in certain data areas.

Dataset diro2c_gaussian_dataset.csv:

Two-dimensional dataset with the continuous features x1 and x2. It was created with µ = 0 and σ 2 = 0.8. It contains 300 instances (data points) with the following properties for feature x1: min = − 275.71, max = 255.90, µ = 0.04, and σ = 88.54 and with the following properties for feature x2: min = − 252.57, max = 201.03, µ = − 9.44, and σ = 86.20. The instances of the datasets are classified into two classes “0” and “1”. 150 are assigned to the class “0” and 150 instances are assigned to the class “1”. The instances of the dataset are generated by the sklearn “make_gaussian_quantiles” function with the following parameters: make_gaussian_quantiles(n_samples = 300, n_classes = 2, shuffle = False, cov = 0.8, random_state = 7). Afterward, we scale the instances by the factor 100.

Dataset diro2c_gaussian_manipulated_dataset.csv:

Two-dimensional dataset with the continuous features x1 and x2. The manipulated dataset is generated with µ = 0 and σ 2 = 1.3. It contains 300 instances (datapoints) with the following properties for feature x1: min = − 351.46, max = 326.21, µ = 0.04, and σ = 112.87 and with the following properties for feature x2: min = − 321.97, max = 256.27, µ = − 12.04, and σ = 109.89. The instances of the datasets are classified into two classes “0” and “1”. 150 instances are assigned to the class “0”, and 150 instances are assigned to the class “1”.