Sampling and Predicting

Used imports:

import bamt.networks as networks
import bamt.preprocessors as pp

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt

from sklearn import preprocessing
from sklearn.metrics import accuracy_score, mean_squared_error
from sklearn.ensemble import RandomForestClassifier
from sklearn.neighbors import KNeighborsClassifier
from sklearn.tree import DecisionTreeClassifier
from sklearn.model_selection import train_test_split

from pgmpy.estimators import K2Score

Importing example data:

data = pd.read_csv(r'../Data/real data/vk_data.csv')
data

Choosing a chunk of data:

cols = ['age', 'sex', 'has_pets', 'is_parent', 'relation', 'is_driver', 'tr_per_month', 'median_tr', 'mean_tr']
data = data[cols]
data[['sex', 'has_pets', 'is_parent', 'relation', 'is_driver']] = data[['sex',      'has_pets',     'is_parent', 'relation',        'is_driver']].astype(str)

Preprocessing data, encode categorical features and discretize numerical features, initialize BN and add nodes:

encoder = preprocessing.LabelEncoder()
discretizer = preprocessing.KBinsDiscretizer(n_bins=5, encode='ordinal', strategy='quantile')

p = pp.Preprocessor([('encoder', encoder), ('discretizer', discretizer)])
discretized_data, est = p.apply(data)

bn = HybridBN(has_logit=True, use_mixture=True) # init BN
info = p.info
info
bn.add_nodes(info)

Learning BN structure with HillClimbing algorithm:

bn.add_edges(discretized_data,  scoring_function=('K2',K2Score))

Learning distribution parameters and sampling data:

bn.fit_parameters(data)
sampled_data = bn.sample(1000) # sample 1000 data points

It is also possible to set where model should stores the data:

bn.sample(1000, models_dir = 'path/to/dir')

Splitting data into train and test sets and dropping target column:

data_test, data_train = train_test_split(data, test_size=0.2, random_state=42)

data_test = data_test.drop(columns=['tr_per_month'])

Predicting the target column with BN:

bn.fit_parameters(data_train)

# parall_count is the number of parallel threads to use
predictions = bn.predict(test=data_test, parall_count=4)