These two datasets were produced to be used in two lectures on Machine Learning for SUSY Model Building taught in pre-SUSY 2023 summer school in Southampton. The code used to generate and to analyse these data can be found here. The datasets are as following: 1 million points generated using SPheno only (so no Dark Matter relic density) for the cMSSM with the physical parameters randomly sampled from the table bellow. The columns are 'm0', 'm12', 'A0', 'tanb': the four physical parameters of the theory 'idx': an utility identifier used during generation, can/should be ignored The flattened SPheno outputs. These are obtained by reading the resulting slha spectrum file outputted by SPheno and flatten the blocks. For example from the 'MINPAR' block, the key-value pairs are given by the columns 'MINPAR_1', 'MINPAR_2', 'MINPAR_3', 'MINPAR_4', 'MINPAR_5', and likewise for all blocks in the slha file. 10 thousand points generated using SPheno, and which spectrum outputs was then fed to micrOMEGAS (MSSM model configured to accept low-scale slha files as input), with the physical parameters randomly sampled from the same table bellow. The columns are: The same as above, in addition to 'Omega', 'dm_spin', 'dm_mass' obtained from the micrOMEGAS output, representing Dark Matter relic density, Dark Matter spin, Dark Matter mass, respectively. The full list of columns can be seen in `column_names.txt` file. Versions: SPheno 4.0.5, with a patch to output a warning when the LSP is charged. This version can be found here. micrOMEGAS 5.3.41, with the MSSM model adapted for low-scale slha inputs. The datasets are provided in Apache `parquet` format. In order to read them using `pandas`, an installation with the optional flag `[parquet]` should be used. Alternatively, one can use `pyarrow`.