def get_train_set(dataset_id):
# download the train set used in training
dt = get_dataset(dataset_id)
filename = __path_filename(dt.filename_train)
ext = filename.split('.')[-1]
return send_file(filename,
as_attachment=True,
attachment_filename='train_%s.%s' % (dataset_id, ext))
def column(dataset_id, col):
# gets the features
dataset = get_dataset(dataset_id)
if col == "None":
return jsonify([
f for f in dataset.features
if f.to_keep and f.name != dataset.y_col
][0].__dict__)
for f in dataset.features:
if f.name == col:
return jsonify(f.__dict__)
return jsonify()
from automlk.dataset import get_dataset
from automlk.graphs import graph_correl_features
for i in range(1, 7):
dt = get_dataset(i)
graph_correl_features(dt)
import pickle
import eli5
import pandas as pd
from eli5.sklearn import PermutationImportance
from automlk.context import get_dataset_folder
from automlk.dataset import get_dataset, get_dataset_sample
dataset_id = '37'
round_id = '19'
dataset = get_dataset(dataset_id)
ds = pickle.load(
open(get_dataset_folder(dataset_id) + '/data/eval_set.pkl', 'rb'))
folder = get_dataset_folder(dataset_id) + '/models/'
names = list(
pickle.load(open(folder + '%s_feature_names.pkl' % round_id, 'rb')))
print(names)
model = pickle.load(open(folder + '%s_model.pkl' % round_id, 'rb'))
pipe_model = pickle.load(open(folder + '%s_pipe_model.pkl' % round_id, 'rb'))
pipe_transform = pickle.load(
open(folder + '%s_pipe_transform.pkl' % round_id, 'rb'))
sample = get_dataset_sample(dataset_id)
X_sample = pipe_transform.transform(pd.DataFrame(sample)[dataset.x_cols])
"""
print('-'*60)
print('test prediction pipeline')
from automlk.store import *
from automlk.dataset import get_dataset_ids, get_dataset, get_dataset_list
print('dataset:counter', get_key_store('dataset:counter'))
print('dataset:list', list_key_store('dataset:list'))
print('dataset ids', get_dataset_ids())
dt = get_dataset(1)
print('dt ok')
l = get_dataset_list()