PiecewiseExponentialFitter seems to be broken

Hi i was running your quickstart example to familiarize with the program and there seems to be an issue generating the broken arrays for the piecewise fitting

from lifelines.datasets import load_waltons
df = load_waltons() # returns a Pandas DataFrame
T = df['T']
E = df['E']

from lifelines import *

fig, axes = plt.subplots(2, 3, figsize=(9, 5))

kmf = KaplanMeierFitter().fit(T, E, label='KaplanMeierFitter')
wbf = WeibullFitter().fit(T, E, label='WeibullFitter')
exf = ExponentialFitter().fit(T, E, label='ExponentalFitter')
lnf = LogNormalFitter().fit(T, E, label='LogNormalFitter')
llf = LogLogisticFitter().fit(T, E, label='LogLogisticFitter')
# pwf = PiecewiseExponentialFitter([40, 60]).fit(T, E, label='PiecewiseExponentialFitter')

wbf.plot_survival_function(ax=axes[0][0])
exf.plot_survival_function(ax=axes[0][1])
lnf.plot_survival_function(ax=axes[0][2])
kmf.plot_survival_function(ax=axes[1][0])
llf.plot_survival_function(ax=axes[1][1])
# pwf.plot_survival_function(ax=axes[1][2])

works, but if I uncomment

 pwf = PiecewiseExponentialFitter([40, 60]).fit(T, E, label='PiecewiseExponentialFitter')

the follwoing error is raised -

Summary: ValueError: cannot reshape array of size 0 into shape (0)

Whole trace -

ValueError                                Traceback (most recent call last)
<ipython-input-9-109e19de1150> in <module>
     29 lnf = LogNormalFitter().fit(T, E, label='LogNormalFitter')
     30 llf = LogLogisticFitter().fit(T, E, label='LogLogisticFitter')
---> 31 pwf = PiecewiseExponentialFitter([40, 60]).fit(T, E, label='PiecewiseExponentialFitter')
     32 
     33 wbf.plot_survival_function(ax=axes[0][0])

/opt/conda/lib/python3.6/site-packages/lifelines/utils/__init__.py in f(self, *args, **kwargs)
     43         def f(self, *args, **kwargs):
     44             self._censoring_type = cls.RIGHT
---> 45             return function(self, *args, **kwargs)
     46 
     47         return f

/opt/conda/lib/python3.6/site-packages/lifelines/fitters/__init__.py in fit(self, durations, event_observed, timeline, label, alpha, ci_labels, show_progress, entry, weights, left_censorship)
    710             show_progress=show_progress,
    711             entry=entry,
--> 712             weights=weights,
    713         )
    714 

/opt/conda/lib/python3.6/site-packages/lifelines/fitters/__init__.py in _fit(self, Ts, event_observed, timeline, label, alpha, ci_labels, show_progress, entry, weights)
    888         # estimation
    889         self._fitted_parameters_, self._log_likelihood, self._hessian_ = self._fit_model(
--> 890             Ts, self.event_observed.astype(bool), self.entry, self.weights, show_progress=show_progress
    891         )
    892 

/opt/conda/lib/python3.6/site-packages/lifelines/fitters/__init__.py in _fit_model(self, Ts, E, entry, weights, show_progress)
    511                 args=(Ts, E, entry, weights),
    512                 bounds=self._bounds,
--> 513                 options={"disp": show_progress},
    514             )
    515 

/opt/conda/lib/python3.6/site-packages/scipy/optimize/_minimize.py in minimize(fun, x0, args, method, jac, hess, hessp, bounds, constraints, tol, callback, options)
    601     elif meth == 'l-bfgs-b':
    602         return _minimize_lbfgsb(fun, x0, args, jac, bounds,
--> 603                                 callback=callback, **options)
    604     elif meth == 'tnc':
    605         return _minimize_tnc(fun, x0, args, jac, bounds, callback=callback,

/opt/conda/lib/python3.6/site-packages/scipy/optimize/lbfgsb.py in _minimize_lbfgsb(fun, x0, args, jac, bounds, disp, maxcor, ftol, gtol, eps, maxfun, maxiter, iprint, callback, maxls, **unknown_options)
    333             # until the completion of the current minimization iteration.
    334             # Overwrite f and g:
--> 335             f, g = func_and_grad(x)
    336         elif task_str.startswith(b'NEW_X'):
    337             # new iteration

/opt/conda/lib/python3.6/site-packages/scipy/optimize/lbfgsb.py in func_and_grad(x)
    283     else:
    284         def func_and_grad(x):
--> 285             f = fun(x, *args)
    286             g = jac(x, *args)
    287             return f, g

/opt/conda/lib/python3.6/site-packages/scipy/optimize/optimize.py in function_wrapper(*wrapper_args)
    291     def function_wrapper(*wrapper_args):
    292         ncalls[0] += 1
--> 293         return function(*(wrapper_args + args))
    294 
    295     return ncalls, function_wrapper

/opt/conda/lib/python3.6/site-packages/scipy/optimize/optimize.py in __call__(self, x, *args)
     61     def __call__(self, x, *args):
     62         self.x = numpy.asarray(x).copy()
---> 63         fg = self.fun(x, *args)
     64         self.jac = fg[1]
     65         return fg[0]

/opt/conda/lib/python3.6/site-packages/autograd/wrap_util.py in nary_f(*args, **kwargs)
     18             else:
     19                 x = tuple(args[i] for i in argnum)
---> 20             return unary_operator(unary_f, x, *nary_op_args, **nary_op_kwargs)
     21         return nary_f
     22     return nary_operator

/opt/conda/lib/python3.6/site-packages/autograd/differential_operators.py in value_and_grad(fun, x)
    128     in scipy.optimize"""
    129     vjp, ans = _make_vjp(fun, x)
--> 130     return ans, vjp(vspace(ans).ones())
    131 
    132 def grad_and_aux(fun, argnum=0):

/opt/conda/lib/python3.6/site-packages/autograd/core.py in vjp(g)
     12         def vjp(g): return vspace(x).zeros()
     13     else:
---> 14         def vjp(g): return backward_pass(g, end_node)
     15     return vjp, end_value
     16 

/opt/conda/lib/python3.6/site-packages/autograd/core.py in backward_pass(g, end_node)
     19     for node in toposort(end_node):
     20         outgrad = outgrads.pop(node)
---> 21         ingrads = node.vjp(outgrad[0])
     22         for parent, ingrad in zip(node.parents, ingrads):
     23             outgrads[parent] = add_outgrads(outgrads.get(parent), ingrad)

/opt/conda/lib/python3.6/site-packages/autograd/core.py in <lambda>(g)
     59                     "VJP of {} wrt argnum 0 not defined".format(fun.__name__))
     60             vjp = vjpfun(ans, *args, **kwargs)
---> 61             return lambda g: (vjp(g),)
     62         elif L == 2:
     63             argnum_0, argnum_1 = argnums

/opt/conda/lib/python3.6/site-packages/autograd/numpy/numpy_vjps.py in <lambda>(g)
    355 def dot_vjp_1(ans, A, B):
    356     A_ndim, B_ndim = anp.ndim(A), anp.ndim(B)
--> 357     return lambda g: dot_adjoint_1(A, g, A_ndim, B_ndim)
    358 defvjp(anp.dot, dot_vjp_0, dot_vjp_1)
    359 

/opt/conda/lib/python3.6/site-packages/autograd/tracer.py in f_wrapped(*args, **kwargs)
     46             return new_box(ans, trace, node)
     47         else:
---> 48             return f_raw(*args, **kwargs)
     49     f_wrapped.fun = f_raw
     50     f_wrapped._is_autograd_primitive = True

/opt/conda/lib/python3.6/site-packages/autograd/numpy/numpy_vjps.py in dot_adjoint_1(A, G, A_ndim, B_ndim)
    347     else:
    348         return swap(onp.tensordot(
--> 349             G, A, [range(-A_ndim - B_ndim + 2, -B_ndim + 1), range(A_ndim - 1)]))
    350 
    351 def dot_vjp_0(ans, A, B):

/opt/conda/lib/python3.6/site-packages/numpy/core/numeric.py in tensordot(a, b, axes)
   1335     oldb = [bs[axis] for axis in notin]
   1336 
-> 1337     at = a.transpose(newaxes_a).reshape(newshape_a)
   1338     bt = b.transpose(newaxes_b).reshape(newshape_b)
   1339     res = dot(at, bt)

ValueError: cannot reshape array of size 0 into shape (0)

[ayadlin]

Upgrading to 1.14 solved he issue!

Thanks-

A

On Mon, Jul 15, 2019 at 11:04 AM Alejandro Wolf Yadlin [e-mail] wrote:

Will do and will let you know!

Thanks so much!

Ale

On Mon, Jul 15, 2019 at 11:03 AM Cameron Davidson-Pilon < [e-mail]> wrote:

Ah, investigating this I have found inconsistent numpy version pinning. In our setup.py, we suggest numpy>=1.6, but in our requirements we suggest numpy>=1.14.0.

The latter is correct, as that’s also what we test against. Can you try upgrading to numpy>=1.14.0?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/CamDavidsonPilon/lifelines/issues/774?email_source=notifications&email_token=ACWHIFYCILRVLYHOIJ3O25DP7S3WPA5CNFSM4ICS5YMKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODZ6P3ZI#issuecomment-511507941, or mute the thread https://github.com/notifications/unsubscribe-auth/ACWHIF3LLO4IX4LLJC2LI7DP7S3WPANCNFSM4ICS5YMA .

– - Alejandro -

– - Alejandro -

[ayadlin]

Will do and will let you know!

Thanks so much!

Ale

On Mon, Jul 15, 2019 at 11:03 AM Cameron Davidson-Pilon < [e-mail]> wrote:

Ah, investigating this I have found inconsistent numpy version pinning. In our setup.py, we suggest numpy>=1.6, but in our requirements we suggest numpy>=1.14.0.

The latter is correct, as that’s also what we test against. Can you try upgrading to numpy>=1.14.0?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/CamDavidsonPilon/lifelines/issues/774?email_source=notifications&email_token=ACWHIFYCILRVLYHOIJ3O25DP7S3WPA5CNFSM4ICS5YMKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODZ6P3ZI#issuecomment-511507941, or mute the thread https://github.com/notifications/unsubscribe-auth/ACWHIF3LLO4IX4LLJC2LI7DP7S3WPANCNFSM4ICS5YMA .

– - Alejandro -