The interpolations package tries to maintain bleeding edge integration with numba and currently supports numba>=0.49. This causes issues with the lecture series dependence on anaconda releases as numba trails in anaconda.
We should see if we can add jitted interpolation functions to QuantEcon.py to replace dependence on interpolations