One of the most well-known libraries for doing statistical analysis is Pandas. This provides a nifty interface to databases basically implemented as a tabular where rows represents different instances in the dataset and columns different properties.
Pandas uses the power of numpy arrays to provide a lot of convenience function to do statistical analysis. It is easy to take average and standard deviation of different quantities, group the dataset by values, select only the data that satisfies certain conditions, and even visualise the data in plots. If we want to give users the ability to do meaningful and easy statistical analysis in a programmatic way, I think that pandas is the way to go.
The downside is that pandas is a bit bulky and adding that dependency is a bit heavy. However, it is easily accessible everywhere (conda and pip).
One of the most well-known libraries for doing statistical analysis is Pandas. This provides a nifty interface to databases basically implemented as a tabular where rows represents different instances in the dataset and columns different properties.
Pandas uses the power of numpy arrays to provide a lot of convenience function to do statistical analysis. It is easy to take average and standard deviation of different quantities, group the dataset by values, select only the data that satisfies certain conditions, and even visualise the data in plots. If we want to give users the ability to do meaningful and easy statistical analysis in a programmatic way, I think that pandas is the way to go.
The downside is that pandas is a bit bulky and adding that dependency is a bit heavy. However, it is easily accessible everywhere (conda and pip).