Aprar
/
bonitoo
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Full set of fields

Browse Source
master
EHP 6 years ago
commit 76645f273d
5 changed files with 995 additions and 0 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Split View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 115
      export.py
  2. 303
      external.ipynb
  3. 2
      src/requirements.txt
  4. 420
      src/train_model.py
  5. 155
      xgboost load.ipynb

115
export.py
Unescape View File

@ -0,0 +1,115 @@
from pymongo import MongoClient
from pprint import pprint
from datetime import datetime
import csv
client = MongoClient()
db=client.bonitoo
fieldnames = [
'timestamp',
'client.channel',
'type',
'flight.inboundSegments.departure',
'flight.inboundSegments.arrival',
'flight.inboundSegments.origin.airportCode',
'flight.inboundSegments.destination.airportCode',
'flight.inboundSegments.flightNumber',
'flight.inboundSegments.travelClass',
'flight.inboundSegments.bookingCode',
'flight.inboundSegments.availability',
'flight.inboundSegments.elapsedFlyingTime',
'flight.outboundSegments.departure',
'flight.outboundSegments.arrival',
'flight.outboundSegments.origin.airportCode',
'flight.outboundSegments.destination.airportCode',
'flight.outboundSegments.flightNumber',
'flight.outboundSegments.travelClass',
'flight.outboundSegments.bookingCode',
'flight.outboundSegments.availability',
'flight.outboundSegments.elapsedFlyingTime',
'flight.inboundEFT', # elapsed flying time
'flight.outboundEFT',
'oneWay',
'adults', # pocet osob = (adults + children)
'children',
'infants',
'input.price',
'input.tax',
'input.currency',
'success',
'status',
'output.price',
'output.tax',
'output.currency',
'duration' # delka volani do nadrazeneho systemu
]
# 5% nebo 200 kc rozdil nahoru
# -200 kc dolu
# abs(+-10kc) ignorovat
# timestamp + ok price - ma byt v cache od cacheat
# timestamp + notok price - nema byt v cache od cacheat
# delka pobytu prilet-odlet
# delka letu ?
# pokud je chyba tak nocache (= chybi priceout)
# brat v uvahu in/out kody aerolinek (mcx ?) - mirek jeste zjisti
# vypocitat uspesnost je/neni v cache v %
counter = 0
with open('export.csv', mode='w') as ef:
writer = csv.DictWriter(ef, fieldnames=fieldnames, delimiter=',', quotechar='"', quoting=csv.QUOTE_MINIMAL)
# do not write header for s3 files
# writer.writeheader()
for it in db.pricing_audit.find():
counter += 1
if counter % 1000 == 0:
print('Iterace %d' % counter)
d = {
'timestamp': datetime.fromtimestamp(it['timestamp']/1000).isoformat(),
'client.channel': it['client']['channel'],
'type': it['type'],
'flight.outboundSegments.departure': '|'.join([x['departure'].isoformat() for x in it['flight']['outboundSegments']]),
'flight.outboundSegments.arrival': '|'.join([x['arrival'].isoformat() for x in it['flight']['outboundSegments']]),
'flight.outboundSegments.origin.airportCode': '|'.join([x['origin']['airportCode'] for x in it['flight']['outboundSegments']]),
'flight.outboundSegments.destination.airportCode': '|'.join([x['destination']['airportCode'] for x in it['flight']['outboundSegments']]),
'flight.outboundSegments.flightNumber': '|'.join([x['flightNumber'] for x in it['flight']['outboundSegments']]),
'flight.outboundSegments.travelClass': '|'.join([x['travelClass'] for x in it['flight']['outboundSegments']]),
'flight.outboundSegments.bookingCode': '|'.join([x.get('bookingCode','') for x in it['flight']['outboundSegments']]),
'flight.outboundSegments.availability': '|'.join([str(x.get('availability','')) for x in it['flight']['outboundSegments']]),
'flight.outboundSegments.elapsedFlyingTime': '|'.join([str(x.get('elapsedFlyingTime','')) for x in it['flight']['outboundSegments']]),
'flight.inboundEFT': it['flight'].get('inboundEFT',''),
'flight.outboundEFT': it['flight'].get('outboundEFT',''),
'oneWay': it['oneWay'],
'adults': it['adults'],
'children': it['children'],
'infants': it['infants'],
'input.price': it['input']['price'],
'input.tax': it['input']['tax'],
'input.currency': it['input']['currency'],
'success': it['success'],
'status': it.get('status',''),
'output.price': it.get('output', {'price': 0})['price'],
'output.tax': it.get('output', {'tax': 0})['tax'],
'output.currency': it.get('output', {'currency': 0})['currency'],
'duration': it['duration']
}
if 'inboundSegments' in it['flight']:
inb = {
'flight.inboundSegments.departure': '|'.join([x['departure'].isoformat() for x in it['flight']['inboundSegments']]),
'flight.inboundSegments.arrival': '|'.join([x['arrival'].isoformat() for x in it['flight']['inboundSegments']]),
'flight.inboundSegments.origin.airportCode': '|'.join([x['origin']['airportCode'] for x in it['flight']['inboundSegments']]),
'flight.inboundSegments.destination.airportCode': '|'.join([x['destination']['airportCode'] for x in it['flight']['inboundSegments']]),
'flight.inboundSegments.flightNumber': '|'.join([x['flightNumber'] for x in it['flight']['inboundSegments']]),
'flight.inboundSegments.travelClass': '|'.join([x['travelClass'] for x in it['flight']['inboundSegments']]),
'flight.inboundSegments.bookingCode': '|'.join([x.get('bookingCode', '') for x in it['flight']['inboundSegments']]),
'flight.inboundSegments.availability': '|'.join([str(x.get('availability','')) for x in it['flight']['inboundSegments']]),
'flight.inboundSegments.elapsedFlyingTime': '|'.join([str(x.get('elapsedFlyingTime','')) for x in it['flight']['inboundSegments']])
}
d = {**d, **inb}
writer.writerow(d)

303
external.ipynb
Unescape View File

@ -0,0 +1,303 @@
{
"cells": [
{
"cell_type": "code",
"execution_count": 1,
"metadata": {},
"outputs": [],
"source": [
"import sagemaker\n",
"import boto3\n",
"from sagemaker import get_execution_role\n",
"\n",
"boto_session = boto3.Session(profile_name='bonitoo', region_name='eu-central-1')\n",
"sagemaker_session = sagemaker.LocalSession(boto_session=boto_session)\n",
"#sagemaker_session = sagemaker.Session()"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [],
"source": [
"# Get a SageMaker-compatible role used by this Notebook Instance.\n",
"#role = get_execution_role()\n",
"role = 'Bonitoo_SageMaker_Execution'\n",
"region = boto_session.region_name"
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {},
"outputs": [],
"source": [
"train_input = 's3://customers-bonitoo-cachettl/sagemaker/data/export.csv'"
]
},
{
"cell_type": "code",
"execution_count": 50,
"metadata": {},
"outputs": [],
"source": [
"from sagemaker.xgboost.estimator import XGBoost\n",
"\n",
"tf = XGBoost(\n",
" entry_point='train_model.py',\n",
" source_dir='./src',\n",
" train_instance_type='local',\n",
" train_instance_count=1,\n",
" role=role,\n",
" sagemaker_session=sagemaker_session,\n",
" framework_version='0.90-1',\n",
" py_version='py3',\n",
" hyperparameters={\n",
" 'bonitoo_price_limit': 1000,\n",
" 'num_round': 15,\n",
" 'max_depth': 15,\n",
" 'eta': 0.5,\n",
" 'num_class': 8,\n",
" 'objective': 'multi:softmax',\n",
" 'eval_metric': 'mlogloss'\n",
" })"
]
},
{
"cell_type": "code",
"execution_count": 51,
"metadata": {
"scrolled": true
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Creating tmptao5hpuc_algo-1-x6dhm_1 ... \n",
"\u001b[1BAttaching to tmptao5hpuc_algo-1-x6dhm_12mdone\u001b[0m\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m INFO:sagemaker-containers:Imported framework sagemaker_xgboost_container.training\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m INFO:sagemaker-containers:No GPUs detected (normal if no gpus installed)\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m INFO:sagemaker_xgboost_container.training:Invoking user training script.\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m INFO:sagemaker-containers:Module train_model does not provide a setup.py. \n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m Generating setup.py\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m INFO:sagemaker-containers:Generating setup.cfg\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m INFO:sagemaker-containers:Generating MANIFEST.in\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m INFO:sagemaker-containers:Installing module with the following command:\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m /usr/bin/python3 -m pip install . -r requirements.txt\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m Processing /opt/ml/code\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m Requirement already satisfied: pandas in /usr/local/lib/python3.5/dist-packages (from -r requirements.txt (line 1)) (0.24.2)\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m Requirement already satisfied: numpy in /usr/local/lib/python3.5/dist-packages (from -r requirements.txt (line 2)) (1.17.2)\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m Requirement already satisfied: pytz>=2011k in /usr/local/lib/python3.5/dist-packages (from pandas->-r requirements.txt (line 1)) (2019.2)\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m Requirement already satisfied: python-dateutil>=2.5.0 in /usr/local/lib/python3.5/dist-packages (from pandas->-r requirements.txt (line 1)) (2.8.0)\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m Requirement already satisfied: six>=1.5 in /usr/local/lib/python3.5/dist-packages (from python-dateutil>=2.5.0->pandas->-r requirements.txt (line 1)) (1.12.0)\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m Building wheels for collected packages: train-model\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m Building wheel for train-model (setup.py) ... \u001b[?25ldone\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \u001b[?25h Created wheel for train-model: filename=train_model-1.0.0-py2.py3-none-any.whl size=6578 sha256=f2f4bac7a2d0260f534e32b3ac0341fb291f30669499adf59ead09aa62b7ccc5\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m Stored in directory: /tmp/pip-ephem-wheel-cache-vdfjugbr/wheels/35/24/16/37574d11bf9bde50616c67372a334f94fa8356bc7164af8ca3\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m Successfully built train-model\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m Installing collected packages: train-model\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m Successfully installed train-model-1.0.0\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m INFO:sagemaker-containers:No GPUs detected (normal if no gpus installed)\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m INFO:sagemaker-containers:Invoking user script\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m Training Env:\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m {\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"network_interface_name\": \"eth0\",\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"hosts\": [\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"algo-1-x6dhm\"\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m ],\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"log_level\": 20,\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"input_config_dir\": \"/opt/ml/input/config\",\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"framework_module\": \"sagemaker_xgboost_container.training:main\",\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"input_dir\": \"/opt/ml/input\",\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"channel_input_dirs\": {\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"training\": \"/opt/ml/input/data/training\"\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m },\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"num_gpus\": 0,\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"job_name\": \"sagemaker-xgboost-2019-10-05-20-16-58-398\",\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"output_intermediate_dir\": \"/opt/ml/output/intermediate\",\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"user_entry_point\": \"train_model.py\",\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"current_host\": \"algo-1-x6dhm\",\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"module_dir\": \"s3://sagemaker-eu-central-1-029917565482/sagemaker-xgboost-2019-10-05-20-16-58-398/source/sourcedir.tar.gz\",\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"master_hostname\": \"algo-1-x6dhm\",\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"module_name\": \"train_model\",\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"resource_config\": {\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"current_host\": \"algo-1-x6dhm\",\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"hosts\": [\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"algo-1-x6dhm\"\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m ]\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m },\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"additional_framework_parameters\": {},\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"num_cpus\": 6,\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"output_data_dir\": \"/opt/ml/output/data\",\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"input_data_config\": {\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"training\": {\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"TrainingInputMode\": \"File\"\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m }\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m },\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"is_master\": true,\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"hyperparameters\": {\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"bonitoo_price_limit\": 1000,\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"max_depth\": 15,\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"objective\": \"multi:softmax\",\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"num_class\": 8,\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"eta\": 0.5,\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"eval_metric\": \"mlogloss\",\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"num_round\": 15\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m },\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"output_dir\": \"/opt/ml/output\",\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"model_dir\": \"/opt/ml/model\"\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m }\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m Environment variables:\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_INPUT_CONFIG_DIR=/opt/ml/input/config\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_HP_MAX_DEPTH=15\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_LOG_LEVEL=20\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_OUTPUT_DIR=/opt/ml/output\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_NUM_CPUS=6\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_CHANNELS=[\"training\"]\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_HP_NUM_ROUND=15\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_HP_OBJECTIVE=multi:softmax\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_OUTPUT_DATA_DIR=/opt/ml/output/data\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_FRAMEWORK_MODULE=sagemaker_xgboost_container.training:main\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_NETWORK_INTERFACE_NAME=eth0\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_INPUT_DATA_CONFIG={\"training\":{\"TrainingInputMode\":\"File\"}}\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_FRAMEWORK_PARAMS={}\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_HPS={\"bonitoo_price_limit\":1000,\"eta\":0.5,\"eval_metric\":\"mlogloss\",\"max_depth\":15,\"num_class\":8,\"num_round\":15,\"objective\":\"multi:softmax\"}\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m PYTHONPATH=/usr/local/bin:/:/usr/local/lib/python3.5/dist-packages/xgboost/dmlc-core/tracker:/usr/lib/python35.zip:/usr/lib/python3.5:/usr/lib/python3.5/plat-x86_64-linux-gnu:/usr/lib/python3.5/lib-dynload:/usr/local/lib/python3.5/dist-packages:/usr/lib/python3/dist-packages\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_RESOURCE_CONFIG={\"current_host\":\"algo-1-x6dhm\",\"hosts\":[\"algo-1-x6dhm\"]}\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_NUM_GPUS=0\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_HP_ETA=0.5\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_HP_NUM_CLASS=8\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_MODULE_DIR=s3://sagemaker-eu-central-1-029917565482/sagemaker-xgboost-2019-10-05-20-16-58-398/source/sourcedir.tar.gz\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_USER_ARGS=[\"--bonitoo_price_limit\",\"1000\",\"--eta\",\"0.5\",\"--eval_metric\",\"mlogloss\",\"--max_depth\",\"15\",\"--num_class\",\"8\",\"--num_round\",\"15\",\"--objective\",\"multi:softmax\"]\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_USER_ENTRY_POINT=train_model.py\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_CURRENT_HOST=algo-1-x6dhm\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_INPUT_DIR=/opt/ml/input\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_CHANNEL_TRAINING=/opt/ml/input/data/training\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_OUTPUT_INTERMEDIATE_DIR=/opt/ml/output/intermediate\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_HP_EVAL_METRIC=mlogloss\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_MODULE_NAME=train_model\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_HP_BONITOO_PRICE_LIMIT=1000\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_HOSTS=[\"algo-1-x6dhm\"]\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_TRAINING_ENV={\"additional_framework_parameters\":{},\"channel_input_dirs\":{\"training\":\"/opt/ml/input/data/training\"},\"current_host\":\"algo-1-x6dhm\",\"framework_module\":\"sagemaker_xgboost_container.training:main\",\"hosts\":[\"algo-1-x6dhm\"],\"hyperparameters\":{\"bonitoo_price_limit\":1000,\"eta\":0.5,\"eval_metric\":\"mlogloss\",\"max_depth\":15,\"num_class\":8,\"num_round\":15,\"objective\":\"multi:softmax\"},\"input_config_dir\":\"/opt/ml/input/config\",\"input_data_config\":{\"training\":{\"TrainingInputMode\":\"File\"}},\"input_dir\":\"/opt/ml/input\",\"is_master\":true,\"job_name\":\"sagemaker-xgboost-2019-10-05-20-16-58-398\",\"log_level\":20,\"master_hostname\":\"algo-1-x6dhm\",\"model_dir\":\"/opt/ml/model\",\"module_dir\":\"s3://sagemaker-eu-central-1-029917565482/sagemaker-xgboost-2019-10-05-20-16-58-398/source/sourcedir.tar.gz\",\"module_name\":\"train_model\",\"network_interface_name\":\"eth0\",\"num_cpus\":6,\"num_gpus\":0,\"output_data_dir\":\"/opt/ml/output/data\",\"output_dir\":\"/opt/ml/output\",\"output_intermediate_dir\":\"/opt/ml/output/intermediate\",\"resource_config\":{\"current_host\":\"algo-1-x6dhm\",\"hosts\":[\"algo-1-x6dhm\"]},\"user_entry_point\":\"train_model.py\"}\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m SM_MODEL_DIR=/opt/ml/model\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m Invoking script with the following command:\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m /usr/bin/python3 -m train_model --bonitoo_price_limit 1000 --eta 0.5 --eval_metric mlogloss --max_depth 15 --num_class 8 --num_round 15 --objective multi:softmax\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m ERROR:sagemaker-containers:ExecuteUserScriptError:\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m Command \"/usr/bin/python3 -m train_model --bonitoo_price_limit 1000 --eta 0.5 --eval_metric mlogloss --max_depth 15 --num_class 8 --num_round 15 --objective multi:softmax\"\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m INFO:root:hyperparameters {'num_round': 15, 'num_class': 8, 'objective': 'multi:softmax', 'eta': 0.5, 'max_depth': 15, 'eval_metric': ['mlogloss']}\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m INFO:root:channels {'training': {'TrainingInputMode': 'File'}}\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m INFO:root:Determined delimiter of CSV input is ','\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m INFO:root:Loading csv file export.csv\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m INFO:root:Preprocessing start\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m /usr/local/lib/python3.5/dist-packages/pandas/core/indexing.py:543: SettingWithCopyWarning: \n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m A value is trying to be set on a copy of a slice from a DataFrame.\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m Try using .loc[row_indexer,col_indexer] = value instead\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m self.obj[item] = s\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m /usr/local/lib/python3.5/dist-packages/pandas/core/indexing.py:362: SettingWithCopyWarning: \n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m A value is trying to be set on a copy of a slice from a DataFrame.\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m Try using .loc[row_indexer,col_indexer] = value instead\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m self.obj[key] = _infer_fill_value(value)\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m INFO:root:Computing cached times\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m INFO:root:Splitting dataset with ration 0.800000\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m /usr/local/lib/python3.5/dist-packages/xgboost/core.py:587: FutureWarning: Series.base is deprecated and will be removed in a future version\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m if getattr(data, 'base', None) is not None and \\\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m /usr/local/lib/python3.5/dist-packages/xgboost/core.py:588: FutureWarning: Series.base is deprecated and will be removed in a future version\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m data.base is not None and isinstance(data, np.ndarray) \\\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m INFO:root:Single node training.\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m INFO:root:Train matrix has 25393 rows\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m INFO:root:Validation matrix has 6314 rows\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m INFO:root:cols: ['index', 'adults', 'children', 'infants', 'input.price', 'input.tax', 'status', 'output.price', 'output.tax', 'duration', 'client.channel_codes', 'type_codes', 'flight.inboundSegments.departure_codes', 'flight.inboundSegments.arrival_codes', 'flight.inboundSegments.origin.airportCode_codes', 'flight.inboundSegments.destination.airportCode_codes', 'flight.inboundSegments.flightNumber_codes', 'flight.inboundSegments.travelClass_codes', 'flight.inboundSegments.bookingCode_codes', 'flight.inboundSegments.availability_codes', 'flight.inboundSegments.elapsedFlyingTime_codes', 'flight.outboundSegments.departure_codes', 'flight.outboundSegments.arrival_codes', 'flight.outboundSegments.origin.airportCode_codes', 'flight.outboundSegments.destination.airportCode_codes', 'flight.outboundSegments.flightNumber_codes', 'flight.outboundSegments.travelClass_codes', 'flight.outboundSegments.bookingCode_codes', 'flight.outboundSegments.availability_codes', 'flight.outboundSegments.elapsedFlyingTime_codes', 'flight.inboundEFT_codes', 'flight.outboundEFT_codes', 'input.currency_codes', 'output.currency_codes', 'oneWay_codes']\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m Traceback (most recent call last):\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m File \"/usr/lib/python3.5/runpy.py\", line 184, in _run_module_as_main\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m \"__main__\", mod_spec)\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m File \"/usr/lib/python3.5/runpy.py\", line 85, in _run_code\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m exec(code, run_globals)\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m File \"/opt/ml/code/train_model.py\", line 417, in <module>\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m checkpoint_config=checkpoint_config\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m File \"/opt/ml/code/train_model.py\", line 320, in sagemaker_train\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m train_job(**train_args)\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m File \"/opt/ml/code/train_model.py\", line 364, in train_job\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m raise Exception(\"cols: %s\" % str(train_dmatrix.feature_names))\n",
"\u001b[36malgo-1-x6dhm_1 |\u001b[0m Exception: cols: ['index', 'adults', 'children', 'infants', 'input.price', 'input.tax', 'status', 'output.price', 'output.tax', 'duration', 'client.channel_codes', 'type_codes', 'flight.inboundSegments.departure_codes', 'flight.inboundSegments.arrival_codes', 'flight.inboundSegments.origin.airportCode_codes', 'flight.inboundSegments.destination.airportCode_codes', 'flight.inboundSegments.flightNumber_codes', 'flight.inboundSegments.travelClass_codes', 'flight.inboundSegments.bookingCode_codes', 'flight.inboundSegments.availability_codes', 'flight.inboundSegments.elapsedFlyingTime_codes', 'flight.outboundSegments.departure_codes', 'flight.outboundSegments.arrival_codes', 'flight.outboundSegments.origin.airportCode_codes', 'flight.outboundSegments.destination.airportCode_codes', 'flight.outboundSegments.flightNumber_codes', 'flight.outboundSegments.travelClass_codes', 'flight.outboundSegments.bookingCode_codes', 'flight.outboundSegments.availability_codes', 'flight.outboundSegments.elapsedFlyingTime_codes', 'flight.inboundEFT_codes', 'flight.outboundEFT_codes', 'input.currency_codes', 'output.currency_codes', 'oneWay_codes']\n",
"\u001b[36mtmptao5hpuc_algo-1-x6dhm_1 exited with code 1\n",
"\u001b[0mAborting on container exit...\n"
]
},
{
"ename": "RuntimeError",
"evalue": "Failed to run: ['docker-compose', '-f', '/tmp/tmptao5hpuc/docker-compose.yaml', 'up', '--build', '--abort-on-container-exit'], Process exited with code: 1",
"output_type": "error",
"traceback": [
"\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
"\u001b[0;31mRuntimeError\u001b[0m Traceback (most recent call last)",
"\u001b[0;32m~/aprar/bonitoo/.venv/lib/python3.7/site-packages/sagemaker/local/image.py\u001b[0m in \u001b[0;36mtrain\u001b[0;34m(self, input_data_config, output_data_config, hyperparameters, job_name)\u001b[0m\n\u001b[1;32m 147\u001b[0m \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 148\u001b[0;31m \u001b[0m_stream_output\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mprocess\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 149\u001b[0m \u001b[0;32mexcept\u001b[0m \u001b[0mRuntimeError\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
"\u001b[0;32m~/aprar/bonitoo/.venv/lib/python3.7/site-packages/sagemaker/local/image.py\u001b[0m in \u001b[0;36m_stream_output\u001b[0;34m(process)\u001b[0m\n\u001b[1;32m 656\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0mexit_code\u001b[0m \u001b[0;34m!=\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 657\u001b[0;31m \u001b[0;32mraise\u001b[0m \u001b[0mRuntimeError\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"Process exited with code: %s\"\u001b[0m \u001b[0;34m%\u001b[0m \u001b[0mexit_code\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 658\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
"\u001b[0;31mRuntimeError\u001b[0m: Process exited with code: 1",
"\nDuring handling of the above exception, another exception occurred:\n",
"\u001b[0;31mRuntimeError\u001b[0m Traceback (most recent call last)",
"\u001b[0;32m<ipython-input-51-ffc1ca8d95fd>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mestimator\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtf\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m{\u001b[0m\u001b[0;34m'training'\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mtrain_input\u001b[0m\u001b[0;34m}\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 2\u001b[0m \u001b[0;31m#estimator = sklearn.attach('sagemaker-scikit-learn-2019-01-25-16-34-38-829')\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
"\u001b[0;32m~/aprar/bonitoo/.venv/lib/python3.7/site-packages/sagemaker/estimator.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, inputs, wait, logs, job_name)\u001b[0m\n\u001b[1;32m 337\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_prepare_for_training\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mjob_name\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mjob_name\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 338\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 339\u001b[0;31m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mlatest_training_job\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_TrainingJob\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstart_new\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minputs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 340\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0mwait\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 341\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mlatest_training_job\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwait\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlogs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mlogs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
"\u001b[0;32m~/aprar/bonitoo/.venv/lib/python3.7/site-packages/sagemaker/estimator.py\u001b[0m in \u001b[0;36mstart_new\u001b[0;34m(cls, estimator, inputs)\u001b[0m\n\u001b[1;32m 861\u001b[0m \u001b[0mcls\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_add_spot_checkpoint_args\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlocal_mode\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mestimator\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtrain_args\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 862\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 863\u001b[0;31m \u001b[0mestimator\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msagemaker_session\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtrain\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m**\u001b[0m\u001b[0mtrain_args\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 864\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 865\u001b[0m \u001b[0;32mreturn\u001b[0m \u001b[0mcls\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mestimator\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msagemaker_session\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mestimator\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_current_job_name\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
"\u001b[0;32m~/aprar/bonitoo/.venv/lib/python3.7/site-packages/sagemaker/session.py\u001b[0m in \u001b[0;36mtrain\u001b[0;34m(self, input_mode, input_config, role, job_name, output_config, resource_config, vpc_config, hyperparameters, stop_condition, tags, metric_definitions, enable_network_isolation, image, algorithm_arn, encrypt_inter_container_traffic, train_use_spot_instances, checkpoint_s3_uri, checkpoint_local_path)\u001b[0m\n\u001b[1;32m 390\u001b[0m \u001b[0mLOGGER\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0minfo\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"Creating training-job with name: %s\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mjob_name\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 391\u001b[0m \u001b[0mLOGGER\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdebug\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"train request: %s\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mjson\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdumps\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtrain_request\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mindent\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m4\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 392\u001b[0;31m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msagemaker_client\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcreate_training_job\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m**\u001b[0m\u001b[0mtrain_request\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 393\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 394\u001b[0m def compile_model(\n",
"\u001b[0;32m~/aprar/bonitoo/.venv/lib/python3.7/site-packages/sagemaker/local/local_session.py\u001b[0m in \u001b[0;36mcreate_training_job\u001b[0;34m(self, TrainingJobName, AlgorithmSpecification, OutputDataConfig, ResourceConfig, InputDataConfig, **kwargs)\u001b[0m\n\u001b[1;32m 99\u001b[0m \u001b[0mtraining_job\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_LocalTrainingJob\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mcontainer\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 100\u001b[0m \u001b[0mhyperparameters\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mkwargs\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m\"HyperParameters\"\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0;34m\"HyperParameters\"\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mkwargs\u001b[0m \u001b[0;32melse\u001b[0m \u001b[0;34m{\u001b[0m\u001b[0;34m}\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 101\u001b[0;31m \u001b[0mtraining_job\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstart\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mInputDataConfig\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mOutputDataConfig\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhyperparameters\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mTrainingJobName\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 102\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 103\u001b[0m \u001b[0mLocalSagemakerClient\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_training_jobs\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mTrainingJobName\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtraining_job\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
"\u001b[0;32m~/aprar/bonitoo/.venv/lib/python3.7/site-packages/sagemaker/local/entities.py\u001b[0m in \u001b[0;36mstart\u001b[0;34m(self, input_data_config, output_data_config, hyperparameters, job_name)\u001b[0m\n\u001b[1;32m 87\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 88\u001b[0m self.model_artifacts = self.container.train(\n\u001b[0;32m---> 89\u001b[0;31m \u001b[0minput_data_config\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0moutput_data_config\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhyperparameters\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mjob_name\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 90\u001b[0m )\n\u001b[1;32m 91\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mend_time\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mdatetime\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdatetime\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mnow\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
"\u001b[0;32m~/aprar/bonitoo/.venv/lib/python3.7/site-packages/sagemaker/local/image.py\u001b[0m in \u001b[0;36mtrain\u001b[0;34m(self, input_data_config, output_data_config, hyperparameters, job_name)\u001b[0m\n\u001b[1;32m 151\u001b[0m \u001b[0;31m# which contains the exit code and append the command line to it.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 152\u001b[0m \u001b[0mmsg\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m\"Failed to run: %s, %s\"\u001b[0m \u001b[0;34m%\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0mcompose_command\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mstr\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 153\u001b[0;31m \u001b[0;32mraise\u001b[0m \u001b[0mRuntimeError\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmsg\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 154\u001b[0m \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 155\u001b[0m \u001b[0martifacts\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mretrieve_artifacts\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mcompose_data\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0moutput_data_config\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mjob_name\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
"\u001b[0;31mRuntimeError\u001b[0m: Failed to run: ['docker-compose', '-f', '/tmp/tmptao5hpuc/docker-compose.yaml', 'up', '--build', '--abort-on-container-exit'], Process exited with code: 1"
]
}
],
"source": [
"estimator = tf.fit({'training': train_input})\n",
"#estimator = sklearn.attach('sagemaker-scikit-learn-2019-01-25-16-34-38-829')"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.7.3"
}
},
"nbformat": 4,
"nbformat_minor": 2
}

2
src/requirements.txt
Unescape View File

@ -0,0 +1,2 @@
pandas
numpy

420
src/train_model.py
Unescape View File

@ -0,0 +1,420 @@
import os
import json
import logging
import argparse
import pandas as pd
import xgboost as xgb
import numpy as np
from sagemaker_algorithm_toolkit import exceptions as exc
from sagemaker_xgboost_container.constants import sm_env_constants
from sagemaker_xgboost_container.data_utils import get_content_type, get_dmatrix, get_size, validate_data_file_path
from sagemaker_xgboost_container import distributed
from sagemaker_xgboost_container import checkpointing
from sagemaker_xgboost_container.algorithm_mode import channel_validation as cv
from sagemaker_xgboost_container.algorithm_mode import hyperparameter_validation as hpv
from sagemaker_xgboost_container.algorithm_mode import metrics as metrics_mod
from sagemaker_xgboost_container.algorithm_mode import train_utils
from sagemaker_xgboost_container.constants.xgb_constants import CUSTOMER_ERRORS
columns = [
'timestamp',
'client.channel',
'type',
'flight.inboundSegments.departure',
'flight.inboundSegments.arrival',
'flight.inboundSegments.origin.airportCode',
'flight.inboundSegments.destination.airportCode',
'flight.inboundSegments.flightNumber',
'flight.inboundSegments.travelClass',
'flight.inboundSegments.bookingCode',
'flight.inboundSegments.availability',
'flight.inboundSegments.elapsedFlyingTime',
'flight.outboundSegments.departure',
'flight.outboundSegments.arrival',
'flight.outboundSegments.origin.airportCode',
'flight.outboundSegments.destination.airportCode',
'flight.outboundSegments.flightNumber',
'flight.outboundSegments.travelClass',
'flight.outboundSegments.bookingCode',
'flight.outboundSegments.availability',
'flight.outboundSegments.elapsedFlyingTime',
'flight.inboundEFT',
'flight.outboundEFT',
'oneWay',
'adults',
'children',
'infants',
'input.price',
'input.tax',
'input.currency',
'success',
'status',
'output.price',
'output.tax',
'output.currency',
'duration'
]
catcolumns = [
'client.channel',
'type',
'flight.inboundSegments.departure',
'flight.inboundSegments.arrival',
'flight.inboundSegments.origin.airportCode',
'flight.inboundSegments.destination.airportCode',
'flight.inboundSegments.flightNumber',
'flight.inboundSegments.travelClass',
'flight.inboundSegments.bookingCode',
'flight.inboundSegments.availability',
'flight.inboundSegments.elapsedFlyingTime',
'flight.outboundSegments.departure',
'flight.outboundSegments.arrival',
'flight.outboundSegments.origin.airportCode',
'flight.outboundSegments.destination.airportCode',
'flight.outboundSegments.flightNumber',
'flight.outboundSegments.travelClass',
'flight.outboundSegments.bookingCode',
'flight.outboundSegments.availability',
'flight.outboundSegments.elapsedFlyingTime',
'flight.inboundEFT',
'flight.outboundEFT',
'input.currency',
'output.currency',
'oneWay'
]
floatcolumns = [
'input.price',
'input.tax',
'output.price',
'output.tax'
]
intcolumns = [
'adults',
'children',
'infants',
'status',
'duration'
]
pkcolumns = [
'flight.inboundSegments.departure',
'flight.inboundSegments.origin.airportCode',
'flight.inboundSegments.destination.airportCode',
'flight.outboundSegments.departure',
'flight.outboundSegments.origin.airportCode',
'flight.outboundSegments.destination.airportCode'
]
def expected_value(row, cachetime_df, price_limit=1000):
# TODO sum tax + price ?
inprice, outprice = row['input.price'], row['output.price']
pricestatus = abs(inprice - outprice)
# TODO correct in cache time ?
timestamps = cachetime_df.loc[row[pkcolumns].fillna(''),'timestamp']
tdiff = timestamps['max'] - timestamps['min']
if tdiff.shape[0] > 0:
incachetime = tdiff[0]
else:
incachetime = np.timedelta64('NaT')
modifier = 0
if pricestatus > price_limit:
modifier = -1
if pricestatus < 0.1:
if not incachetime:
return 3
modifier = 1
if pd.isnull(incachetime):
return 0
if incachetime <= np.timedelta64(12,'h'):
return 1 + modifier
if incachetime <= np.timedelta64(1,'D'):
return 2 + modifier
if incachetime <= np.timedelta64(2,'D'):
return 3 + modifier
if incachetime <= np.timedelta64(3,'D'):
return 4 + modifier
if incachetime <= np.timedelta64(7,'D'):
return 5 + modifier
if incachetime <= np.timedelta64(14,'D'):
return 6 + modifier
return min(7, 7 + modifier)
def compute_cached_time(df):
logging.info('Computing cached times')
return df.set_index(pkcolumns).groupby(pkcolumns).agg({'timestamp': ['min', 'max']})
def preprocess_data(df):
logging.info('Preprocessing start')
df = df[df.loc[:, 'success'] == True]
df.loc[:, 'timestamp'] = df.loc[:, 'timestamp'].apply(lambda x: pd.to_datetime(x))
booleanDictionary = {True: 'TRUE', False: 'FALSE'}
df.loc[:, 'oneWay'] = df.loc[:, 'oneWay'].replace(booleanDictionary)
for cc in catcolumns:
df.loc[:, cc] = df.loc[:, cc].astype('category')
df.loc[:, '%s_codes' % cc] = df[cc].cat.codes
df.loc[:, floatcolumns] = df.loc[:, floatcolumns].astype('float64')
df.loc[:, intcolumns] = df.loc[:, intcolumns].fillna(-1).astype('int32')
return df
def remove_non_features(df):
return df.drop(['timestamp', 'success'] + catcolumns, axis=1), df
def train_test_split(df, label, ratio):
logging.info('Splitting dataset with ration %f', ratio)
msk = np.random.rand(len(df)) < ratio
train_data = df[msk]
test_data = df[~msk]
train_label = label[msk]
test_label = label[~msk]
train_data = train_data.reset_index()
test_data = test_data.reset_index()
train_label = train_label.reset_index()
test_label = test_label.reset_index()
return train_data, test_data, train_label, test_label
def get_csv_pandas(files_path):
csv_file = files_path if os.path.isfile(files_path) else [
f for f in os.listdir(files_path) if os.path.isfile(os.path.join(files_path, f))][0]
try:
logging.info('Loading csv file %s', csv_file)
df = pd.read_csv(os.path.join(files_path, csv_file), header=None)
df.columns = columns
#raise "cols: " + df.info()
return df
except Exception as e:
raise exc.UserError("Failed to load csv data with exception:\n{}".format(e))
def get_pandas_df(data_path):
if not os.path.exists(data_path):
return None
else:
if os.path.isfile(data_path):
files_path = data_path
else:
for root, dirs, files in os.walk(data_path):
if dirs == []:
files_path = root
break
df = get_csv_pandas(files_path)
return df
def get_df(train_path, validate_path, content_type='text/csv'):
train_files_size = get_size(train_path) if train_path else 0
val_files_size = get_size(validate_path) if validate_path else 0
logging.debug("File size need to be processed in the node: {}mb.".format(
round((train_files_size + val_files_size) / (1024 * 1024), 2)))
if train_files_size > 0:
validate_data_file_path(train_path, content_type)
if val_files_size > 0:
validate_data_file_path(validate_path, content_type)
train_pandas = get_pandas_df(train_path) if train_files_size > 0 else None
val_pandas = get_pandas_df(validate_path) if val_files_size > 0 else None
return train_pandas, val_pandas
def get_dmatrices(train_pandas, train_label_pandas, val_pandas, val_label_pandas, ratio=0.8):
if val_pandas:
train_dmatrix = xgb.DMatrix(train_pandas, label=train_label_pandas.loc[:, 'label'])
val_dmatrix = xgb.DMatrix(val_pandas, label=val_label_pandas.loc[:, 'label'])
else:
train_data, test_data, train_label, test_label = train_test_split(train_pandas, train_label_pandas, ratio)
train_dmatrix = xgb.DMatrix(train_data, label=train_label.loc[:, 'label'])
val_dmatrix = xgb.DMatrix(test_data, label=test_label.loc[:, 'label'])
return train_dmatrix, val_dmatrix
def save_encoders(encoder_location, df):
logging.info('Saving encoders')
jsondata = {}
for cc in catcolumns:
jsondata[cc] = {cat: idx for idx, cat in enumerate(df[cc].cat.categories)}
with open(encoder_location, 'w') as f:
json.dump(jsondata, f)
def sagemaker_train(train_config, data_config, train_path, val_path, model_dir, sm_hosts, sm_current_host,
checkpoint_config):
metrics = metrics_mod.initialize()
hyperparameters = hpv.initialize(metrics)
price_limit = int(train_config.get('bonitoo_price_limit', 1000))
train_config = {k:v.replace('"', '') for k,v in train_config.items() if not k.startswith('sagemaker_') and not k.startswith('bonitoo_')}
train_config = hyperparameters.validate(train_config)
if train_config.get("updater"):
train_config["updater"] = ",".join(train_config["updater"])
logging.info("hyperparameters {}".format(train_config))
logging.info("channels {}".format(data_config))
# Get Training and Validation Data Matrices
validation_channel = data_config.get('validation', None)
checkpoint_dir = checkpoint_config.get("LocalPath", None)
train_df, val_df = get_df(train_path, val_path)
train_df = preprocess_data(train_df)
cachetime_df = compute_cached_time(train_df)
train_label_df = train_df.apply(lambda x: expected_value(x, cachetime_df, price_limit), axis=1).to_frame(name='label')
train_df, train_df_orig = remove_non_features(train_df)
val_label_df = None
if val_df:
val_df = preprocess_data(val_df)
val_label_df = val_df.apply(lambda x: expected_value(x, cachetime_df, price_limit), axis=1).to_frame(name='label')
val_df, val_df_orig = remove_non_features(val_df)
train_dmatrix, val_dmatrix = get_dmatrices(train_df, train_label_df, val_df, val_label_df)
train_args = dict(
train_cfg=train_config,
train_dmatrix=train_dmatrix,
train_df=train_df_orig,
val_dmatrix=val_dmatrix,
model_dir=model_dir,
checkpoint_dir=checkpoint_dir)
# Obtain information about training resources to determine whether to set up Rabit or not
num_hosts = len(sm_hosts)
if num_hosts > 1:
# Wait for hosts to find each other
logging.info("Distributed node training with {} hosts: {}".format(num_hosts, sm_hosts))
distributed.wait_hostname_resolution(sm_hosts)
if not train_dmatrix:
logging.warning("Host {} does not have data. Will broadcast to cluster and will not be used in distributed"
" training.".format(sm_current_host))
distributed.rabit_run(exec_fun=train_job, args=train_args, include_in_training=(train_dmatrix is not None),
hosts=sm_hosts, current_host=sm_current_host, update_rabit_args=True)
elif num_hosts == 1:
if train_dmatrix:
if validation_channel:
if not val_dmatrix:
raise exc.UserError("No data in validation channel path {}".format(val_path))
logging.info("Single node training.")
train_args.update({'is_master': True})
train_job(**train_args)
else:
raise exc.UserError("No data in training channel path {}".format(train_path))
else:
raise exc.PlatformError("Number of hosts should be an int greater than or equal to 1")
def train_job(train_cfg, train_dmatrix, val_dmatrix, train_df, model_dir, checkpoint_dir, is_master):
# Parse arguments for train() API
early_stopping_rounds = train_cfg.get('early_stopping_rounds')
num_round = int(train_cfg["num_round"])
# Evaluation metrics to use with train() API
tuning_objective_metric_param = train_cfg.get("_tuning_objective_metric")
eval_metric = train_cfg.get("eval_metric")
cleaned_eval_metric, configured_feval = train_utils.get_eval_metrics_and_feval(
tuning_objective_metric_param, eval_metric)
if cleaned_eval_metric:
train_cfg['eval_metric'] = cleaned_eval_metric
else:
train_cfg.pop('eval_metric', None)
# Set callback evals
watchlist = [(train_dmatrix, 'train')]
if val_dmatrix is not None:
watchlist.append((val_dmatrix, 'validation'))
xgb_model, iteration = checkpointing.load_checkpoint(checkpoint_dir)
num_round -= iteration
if xgb_model is not None:
logging.info("Checkpoint loaded from %s", xgb_model)
logging.info("Resuming from iteration %s", iteration)
callbacks = []
callbacks.append(checkpointing.print_checkpointed_evaluation(start_iteration=iteration))
if checkpoint_dir:
save_checkpoint = checkpointing.save_checkpoint(checkpoint_dir, start_iteration=iteration)
callbacks.append(save_checkpoint)
logging.info("Train matrix has {} rows".format(train_dmatrix.num_row()))
if val_dmatrix:
logging.info("Validation matrix has {} rows".format(val_dmatrix.num_row()))
# TODO remove
#logging.info("cols: %s", str(train_dmatrix.feature_names))
#raise Exception("cols: %s" % str(train_dmatrix.feature_names))
try:
bst = xgb.train(train_cfg, train_dmatrix, num_boost_round=num_round, evals=watchlist, feval=configured_feval,
early_stopping_rounds=early_stopping_rounds, callbacks=callbacks, xgb_model=xgb_model,
verbose_eval=False)
except Exception as e:
for customer_error_message in CUSTOMER_ERRORS:
if customer_error_message in str(e):
raise exc.UserError(str(e))
exception_prefix = "XGB train call failed with exception"
raise exc.AlgorithmError("{}:\n {}".format(exception_prefix, str(e)))
if not os.path.exists(model_dir):
os.makedirs(model_dir)
if is_master:
encoder_location = model_dir + '/encoder.json'
save_encoders(encoder_location, train_df)
logging.info("Stored encoders at {}".format(encoder_location))
model_location = model_dir + '/xgboost-model.bin'
bst.save_model(model_location)
logging.info("Stored trained model at {}".format(model_location))
if __name__ == '__main__':
with open(os.getenv(sm_env_constants.SM_INPUT_TRAINING_CONFIG_FILE), "r") as f:
train_config = json.load(f)
with open(os.getenv(sm_env_constants.SM_INPUT_DATA_CONFIG_FILE), "r") as f:
data_config = json.load(f)
checkpoint_config_file = os.getenv(sm_env_constants.SM_CHECKPOINT_CONFIG_FILE)
if os.path.exists(checkpoint_config_file):
with open(checkpoint_config_file, "r") as f:
checkpoint_config = json.load(f)
else:
checkpoint_config = {}
train_path = os.environ['SM_CHANNEL_TRAINING']
val_path = os.environ.get(sm_env_constants.SM_CHANNEL_VALIDATION)
sm_hosts = json.loads(os.environ[sm_env_constants.SM_HOSTS])
sm_current_host = os.environ[sm_env_constants.SM_CURRENT_HOST]
model_dir = os.getenv(sm_env_constants.SM_MODEL_DIR)
sagemaker_train(
train_config=train_config, data_config=data_config,
train_path=train_path, val_path=val_path, model_dir=model_dir,
sm_hosts=sm_hosts, sm_current_host=sm_current_host,
checkpoint_config=checkpoint_config
)

155
xgboost load.ipynb
View File

File diff suppressed because one or more lines are too long
Write Preview
Loading…
Cancel
Save