def calc_mode_manual(self, loc=0.0, scale=1.0):
shape = np.array(self.df).shape
dfmax = np.array(self.df * np.ones((1,)), dtype='i')
dfmax = (dfmax.flatten() - 2).tolist()
for id in range(len(dfmax)):
dfmax[id] = np.max([dfmax[id], 0])
return np.reshape(dfmax, shape) + make_int(np.around(loc))
def __init__(self, **params):
self.type = ProbabilityDistributionTypes.CHISQUARE
self.scipy_name = "chi"
self.numpy_name = ProbabilityDistributionTypes.CHISQUARE
self.constraint_string = "int(df) > 0"
self.df = make_int(params.get("df", 1))
self.__update_params__(df=self.df)
def __init__(self, **params):
self.type = ProbabilityDistributionTypes.BINOMIAL
self.scipy_name = "binom"
self.numpy_name = ProbabilityDistributionTypes.BINOMIAL
self.constraint_string = "n > 0 and 0 < p < 1"
self.n = make_int(params.get("n", 1))
self.p = make_float(params.get("p", 0.5))
self.__update_params__(n=self.n, p=self.p)
def calc_mode_manual(self, loc=0.0, scale=1.0):
loc = make_int(np.round(loc))
p = np.array(self.p)
shape = p.shape
mode = np.ones(np.array(p * np.ones((1,))).shape, dtype="i") + loc
mode[np.where(p < 0.5)[0]] = 1 + loc
p05 = p == 0.5
if np.any(p05):
self.logger.warning("The mode of bernoulli distribution for p=0.5 consists of two values (0.0 + loc and 1.0 + loc)!")
mode = mode.astype('O')
mode[np.where(p05)[0]] = [[0.0 + loc, 1.0 + loc]]
return mode
def calc_mean_manual(self, loc=0.0, scale=1.0):
return self.df + make_int(np.around(loc))
def update_params(self, loc=0.0, scale=1.0, use="scipy", **params):
self.__update_params__(loc, scale, use,
df=make_int(params.get("df", self.df)))
def calc_mode_manual(self, loc=0.0, scale=1.0):
return make_int(np.round((self.n + 1) * self.p + loc) - 1)
def calc_median_manual(self, loc=0.0, scale=1.0):
return make_int(np.round(self.calc_mean_manual(loc=loc)))
def update_params(self, loc=0.0, scale=1.0, use="scipy", **params):
self.__update_params__(loc,
scale,
use,
n=make_int(params.get("n", self.n)),
p=make_float(params.get("p", self.p)))
def calc_mode_manual(self, loc=0.0, scale=1.0):
return [
make_int(np.round(self.lamda + loc)) - 1,
make_int(np.round(self.lamda + loc))
]