[Python] Python catboost custom loss function

I know that there is built-in Poisson loss function in catboost. I want to try to reproduce it. I used the tutorial example https://github.com/catboost/catboos...om_loss/custom_loss_and_metric_tutorial.ipynb

My code for CustomObjective and CustomMetric:

class CustomPoissonObjective(object):

def calc_ders_range(self, approxes, targets, weights):

assert len(approxes) == len(targets)
if weights is not None:
assert len(weights) == len(approxes)

result = []
for index in range(len(targets)):
e = np.exp(approxes[index])
der1 = e - targets[index]
der2 = e

if weights is not None:
der1 *= weights[index]
der2 *= weights[index]

result.append((der1, der2))
return result


class CustomPoissonMetric(object):

def is_max_optimal(self):
return False

def evaluate(self, approxes, target, weight):
assert len(approxes) == 1
assert len(target) == len(approxes[0])

approx = approxes[0]

error_sum = 0.0
weight_sum = 0.0
for i in range(len(approx)):
w = 1.0 if weight is None else weight
weight_sum += w
error_sum += w * (np.exp(approx) - target * approx)

return error_sum, weight_sum

def get_final_error(self, error, weight):
return error / (weight + 1e-38)


When trying to reproduce the experiment with built-in loss_fuction='Poisson' and loss_fuction=CustomPoissonObjective() I get different results.

I don't understand where I'm going wrong. I would really appreciate your help!

I've tried CatBoostRegressor(loss_function=CustomPoissonObjective())

Continue reading...