Bin one-dimensional data and calculate conditional expectation values.
The main function is bin_df, which accepts numpy-arrays or pandas Dataframes
as inputs. Bins in linear, logarithmic, or quantile space are created for the
independent variable. The dependent variable is then aggregated in these bins.
Lots of options allow to customise the binning and aggregation, calculate errors, etc.
%pylab
from binny import bin_df
x = randn(10**4)
y = x**2 + randn(10**4)
bin_df(y, by=x, bins=11).plot(marker='x')
# now plot the same data using bins containing an equal number of events
# (quantile bins)
bin_df(y, by=x, bins=11, space='q').plot(marker='+')
More examples can be found in the notebooks directory. Use nbviewer if GitHub fails to render.
pip install binny
- Python >= 2.7 or >= 3.6
- NumPy
- Pandas