vitords / hoeffdingtree Goto Github PK

File "d:\classification_rules_big_data\code\HoeffdingTree-master\main.py", line 2, in
from hoeffdingtree import *
File "d:\classification_rules_big_data\code\HoeffdingTree-master\hoeffdingtree.py", line 5, in
from core.attribute import Attribute
File "d:\classification_rules_big_data\code\HoeffdingTree-master\core\attribute.py", line 3, in
class Attribute(object):
File "d:\classification_rules_big_data\code\HoeffdingTree-master\core\attribute.py", line 151, in Attribute
def set_numeric_range(self, lower_bound=-math.inf, upper_bound=math.inf):

AttributeError: 'module' object has no attribute 'inf'

Hi,sir

I want know, what used data set for train and test, I need use for experiments.
I want to help,what was type data set used code.

Thanks .
Mohammadi

Hi,Dear
it is output for HIGGS dataset, do you know correct by output?
Do you know what I can be doing wrong?

Thanks in advance.
_str nomodel
(11010001.0)
Dataset 'New dataset'
Attributes: ['0', '1', '2', '3', '4', '5', '6', '7', '8', '9', '10', '11', '12', '13', '14', '15', '16', '17', '18', '19', '20', '21', '22', '23', '24', '25', '26', '27', '28']
Class attribute: 0
Total instances: 10000
(11010001.0)
Instance
From dataset: New dataset
Attribute values: ['0.00E+000', '5.47E-001 ', '-3.50E-001', '-6.47E-001', '2.04E+000', '2.76E-001', '5.45E-001', '8.39E-001', '1.73E+000', '0.00E+000', '6.53E-001', '1.47E+000', '1.24E+000', '0.00E+000', '7.86E-001', '-4.44E-002', '-1.02E+000', '2.55E+000', '4.19E-001', '-6.29E-001', '1.57E+000', '3.10E+000', '6.89E-001', '8.67E-001', '1.08E+000', '6.64E-001', '3.54E-001', '5.80E-001', '8.17E-001']
Class: 0.00E+000
[ ]

Process finished with exit code 0

Hi vitords

I used to attached files for experiments but can`t get tree model.
I have another question: how to calculate "pre_dist" and "post_dist" properly for my dataset?

main.txt
dataset and main.txt send email`s
dataset_file.txt

File "d:\A_start_oct30_2016\moa\HoeffdingTree-master\main.py", line 106, in
main()
File "d:\A_start_oct30_2016\moa\HoeffdingTree-master\main.py", line 79, in main
vfdt.set_set_hoeffding_tie_threshold(0.05)

builtins.AttributeError: 'HoeffdingTree' object has no attribute 'set_set_hoeffding_tie_threshold'

it give error
builtins.FileNotFoundError: [Errno 2] No such file or directory: 'dataset_file.csv'
so it would be nice to have data file and results of classification to make sure it runs corectly
for example you may use the same data as in
https://github.com/blazs/HoeffdingTree
housing.dat
housing-100K.dat
sample.dat
sea.dat
winnow.dat

Dear vitords,

I have been using your VFDT (thanks for your great work), and I have found that when I have a new instance and I want to know the predicted class, the method distribution_for_instance should provides the class probabilities for this new instance. But, I have found that the class probabilities are always distributed proportionally to the number of values ​​of the class.

For example: if my class have these values [0,1], the method always gives [0.5,0.5]. If my class have these values [0,1,2], the method always gives [0.33,0.33,0.33], etc.

Do you know what I can be doing wrong?

Thanks in advance.

Hello, :-)
file ht/gaussianconditionalsufficientstats.py:
in the function "probability_density(self, value)", the code follows:

return (1.0 / (self.CONST * std_dev)) * math.exp(-(diff * diff / (2.0 * self._variance))) 

I think this code means the probability density function of Gaussian distribution, but the self.CONST here actually is math.log(2 * math.pi), should it be math.sqrt(2 * math.pi)?
please point it out if I am wrong, thanks