adelorenzo/stable
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

improve data persistence/mapping for live/dry. This accommodates quick reloads after crash and handles multi-pair cleanly

Browse Source
This commit is contained in:
robcaulk 2022-05-23 21:05:05 +02:00
parent e1c068ca66
commit b0d2d13eb1
4 changed files with 199 additions and 131 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

59
freqtrade/freqai/data_drawer.py Normal file
View File

@ -0,0 +1,59 @@
import json
import logging
from pathlib import Path
from typing import Any, Dict, Tuple
# import pickle as pk
import numpy as np
logger = logging.getLogger(__name__)
class FreqaiDataDrawer:
"""
Class aimed at holding all pair models/info in memory for better inferencing/retrainig/saving
/loading to/from disk.
This object remains persistent throughout live/dry, unlike FreqaiDataKitchen, which is
reinstantiated for each coin.
"""
def __init__(self, full_path: Path):
# dictionary holding all pair metadata necessary to load in from disk
self.pair_dict: Dict[str, Any] = {}
# dictionary holding all actively inferenced models in memory given a model filename
self.model_dictionary: Dict[str, Any] = {}
self.full_path = full_path
self.load_drawer_from_disk()
def load_drawer_from_disk(self):
exists = Path(self.full_path / str('pair_dictionary.json')).resolve().exists()
if exists:
with open(self.full_path / str('pair_dictionary.json'), "r") as fp:
self.pair_dict = json.load(fp)
else:
logger.info("Could not find existing datadrawer, starting from scratch")
return exists
def save_drawer_to_disk(self):
with open(self.full_path / str('pair_dictionary.json'), "w") as fp:
json.dump(self.pair_dict, fp, default=self.np_encoder)
def np_encoder(self, object):
if isinstance(object, np.generic):
return object.item()
def get_pair_dict_info(self, metadata: dict) -> Tuple[str, int, bool]:
pair_in_dict = self.pair_dict.get(metadata['pair'])
if pair_in_dict:
model_filename = self.pair_dict[metadata['pair']]['model_filename']
trained_timestamp = self.pair_dict[metadata['pair']]['trained_timestamp']
coin_first = self.pair_dict[metadata['pair']]['first']
else:
self.pair_dict[metadata['pair']] = {}
model_filename = self.pair_dict[metadata['pair']]['model_filename'] = ''
coin_first = self.pair_dict[metadata['pair']]['first'] = True
trained_timestamp = self.pair_dict[metadata['pair']]['trained_timestamp'] = 0
return model_filename, trained_timestamp, coin_first

123
freqtrade/freqai/data_kitchen.py
View File

@ -19,6 +19,7 @@ from sklearn.model_selection import train_test_split
from freqtrade.configuration import TimeRange from freqtrade.configuration import TimeRange
from freqtrade.data.history import load_pair_history from freqtrade.data.history import load_pair_history
from freqtrade.data.history.history_utils import refresh_backtest_ohlcv_data from freqtrade.data.history.history_utils import refresh_backtest_ohlcv_data
from freqtrade.freqai.data_drawer import FreqaiDataDrawer
from freqtrade.resolvers import ExchangeResolver from freqtrade.resolvers import ExchangeResolver
from freqtrade.strategy.interface import IStrategy from freqtrade.strategy.interface import IStrategy
@ -33,13 +34,13 @@ logger = logging.getLogger(__name__)
class FreqaiDataKitchen: class FreqaiDataKitchen:
""" """
Class designed to handle all the data for the IFreqaiModel class model. Class designed to analyze data for a single pair. Employed by the IFreqaiModel class.
Functionalities include holding, saving, loading, and analyzing the data. Functionalities include holding, saving, loading, and analyzing the data.
author: Robert Caulk, rob.caulk@gmail.com author: Robert Caulk, rob.caulk@gmail.com
""" """
def __init__(self, config: Dict[str, Any], dataframe: DataFrame, live: bool = False): def __init__(self, config: Dict[str, Any], data_drawer: FreqaiDataDrawer, live: bool = False,
self.full_dataframe = dataframe pair: str = ''):
self.data: Dict[Any, Any] = {} self.data: Dict[Any, Any] = {}
self.data_dictionary: Dict[Any, Any] = {} self.data_dictionary: Dict[Any, Any] = {}
self.config = config self.config = config
@ -53,10 +54,10 @@ class FreqaiDataKitchen:
self.full_do_predict: npt.ArrayLike = np.array([]) self.full_do_predict: npt.ArrayLike = np.array([])
self.full_target_mean: npt.ArrayLike = np.array([]) self.full_target_mean: npt.ArrayLike = np.array([])
self.full_target_std: npt.ArrayLike = np.array([]) self.full_target_std: npt.ArrayLike = np.array([])
self.model_path = Path() self.data_path = Path()
self.model_filename: str = "" self.model_filename: str = ""
self.model_dictionary: Dict[Any, Any] = {}
self.live = live self.live = live
self.pair = pair
self.svm_model: linear_model.SGDOneClassSVM = None self.svm_model: linear_model.SGDOneClassSVM = None
if not self.live: if not self.live:
self.full_timerange = self.create_fulltimerange(self.config["timerange"], self.full_timerange = self.create_fulltimerange(self.config["timerange"],
@ -69,6 +70,8 @@ class FreqaiDataKitchen:
config["freqai"]["backtest_period"], config["freqai"]["backtest_period"],
) )
self.data_drawer = data_drawer
def assert_config(self, config: Dict[str, Any], live: bool) -> None: def assert_config(self, config: Dict[str, Any], live: bool) -> None:
assert config.get('freqai'), "No Freqai parameters found in config file." assert config.get('freqai'), "No Freqai parameters found in config file."
assert config.get('freqai', {}).get('train_period'), ("No Freqai train_period found in" assert config.get('freqai', {}).get('train_period'), ("No Freqai train_period found in"
@ -88,18 +91,18 @@ class FreqaiDataKitchen:
assert config.get('freqai', {}).get('feature_parameters'), ("No Freqai feature_parameters" assert config.get('freqai', {}).get('feature_parameters'), ("No Freqai feature_parameters"
"found in config file.") "found in config file.")
def set_paths(self) -> None: def set_paths(self, trained_timestamp: int = None) -> None:
self.full_path = Path(self.config['user_data_dir'] / self.full_path = Path(self.config['user_data_dir'] /
"models" / "models" /
str(self.freqai_config.get('live_full_backtestrange') + str(self.freqai_config.get('live_full_backtestrange') +
self.freqai_config.get('identifier'))) self.freqai_config.get('identifier')))
self.model_path = Path(self.full_path / str("sub-train" + "-" + self.data_path = Path(self.full_path / str("sub-train" + "-" + self.pair.split("/")[0] +
str(self.freqai_config.get('live_trained_timerange')))) str(trained_timestamp)))
return return
def save_data(self, model: Any) -> None: def save_data(self, model: Any, coin: str = '') -> None:
""" """
Saves all data associated with a model for a single sub-train time range Saves all data associated with a model for a single sub-train time range
:params: :params:
@ -107,10 +110,10 @@ class FreqaiDataKitchen:
predictions predictions
""" """
if not self.model_path.is_dir(): if not self.data_path.is_dir():
self.model_path.mkdir(parents=True, exist_ok=True) self.data_path.mkdir(parents=True, exist_ok=True)
save_path = Path(self.model_path) save_path = Path(self.data_path)
# Save the trained model # Save the trained model
dump(model, save_path / str(self.model_filename + "_model.joblib")) dump(model, save_path / str(self.model_filename + "_model.joblib"))
@ -118,7 +121,7 @@ class FreqaiDataKitchen:
if self.svm_model is not None: if self.svm_model is not None:
dump(self.svm_model, save_path / str(self.model_filename + "_svm_model.joblib")) dump(self.svm_model, save_path / str(self.model_filename + "_svm_model.joblib"))
self.data["model_path"] = str(self.model_path) self.data["data_path"] = str(self.data_path)
self.data["model_filename"] = str(self.model_filename) self.data["model_filename"] = str(self.model_filename)
self.data["training_features_list"] = list(self.data_dictionary["train_features"].columns) self.data["training_features_list"] = list(self.data_dictionary["train_features"].columns)
# store the metadata # store the metadata
@ -131,7 +134,10 @@ class FreqaiDataKitchen:
) )
if self.live: if self.live:
self.model_dictionary[self.model_filename] = model self.data_drawer.model_dictionary[self.model_filename] = model
self.data_drawer.pair_dict[coin]['model_filename'] = self.model_filename
self.data_drawer.pair_dict[coin]['data_path'] = str(self.data_path)
self.data_drawer.save_drawer_to_disk()
# TODO add a helper function to let user save/load any data they are custom adding. We # TODO add a helper function to let user save/load any data they are custom adding. We
# do not want them having to edit the default save/load methods here. Below is an example # do not want them having to edit the default save/load methods here. Below is an example
@ -148,19 +154,23 @@ class FreqaiDataKitchen:
return return
def load_data(self) -> Any: def load_data(self, coin: str = '') -> Any:
""" """
loads all data required to make a prediction on a sub-train time range loads all data required to make a prediction on a sub-train time range
:returns: :returns:
:model: User trained model which can be inferenced for new predictions :model: User trained model which can be inferenced for new predictions
""" """
with open(self.model_path / str(self.model_filename + "_metadata.json"), "r") as fp: if self.live:
self.model_filename = self.data_drawer.pair_dict[coin]['model_filename']
self.data_path = Path(self.data_drawer.pair_dict[coin]['data_path'])
with open(self.data_path / str(self.model_filename + "_metadata.json"), "r") as fp:
self.data = json.load(fp) self.data = json.load(fp)
self.training_features_list = self.data["training_features_list"] self.training_features_list = self.data["training_features_list"]
self.data_dictionary["train_features"] = pd.read_pickle( self.data_dictionary["train_features"] = pd.read_pickle(
self.model_path / str(self.model_filename + "_trained_df.pkl") self.data_path / str(self.model_filename + "_trained_df.pkl")
) )
# TODO add a helper function to let user save/load any data they are custom adding. We # TODO add a helper function to let user save/load any data they are custom adding. We
@ -169,34 +179,34 @@ class FreqaiDataKitchen:
# if self.freqai_config.get('feature_parameters','determine_statistical_distributions'): # if self.freqai_config.get('feature_parameters','determine_statistical_distributions'):
# self.data_dictionary["upper_quantiles"] = pd.read_pickle( # self.data_dictionary["upper_quantiles"] = pd.read_pickle(
# self.model_path / str(self.model_filename + "_upper_quantiles.pkl") # self.data_path / str(self.model_filename + "_upper_quantiles.pkl")
# ) # )
# self.data_dictionary["lower_quantiles"] = pd.read_pickle( # self.data_dictionary["lower_quantiles"] = pd.read_pickle(
# self.model_path / str(self.model_filename + "_lower_quantiles.pkl") # self.data_path / str(self.model_filename + "_lower_quantiles.pkl")
# ) # )
self.model_path = Path(self.data["model_path"]) # self.data_path = Path(self.data["data_path"])
self.model_filename = self.data["model_filename"] # self.model_filename = self.data["model_filename"]
# try to access model in memory instead of loading object from disk to save time # try to access model in memory instead of loading object from disk to save time
if self.live and self.model_filename in self.model_dictionary: if self.live and self.model_filename in self.data_drawer.model_dictionary:
model = self.model_dictionary[self.model_filename] model = self.data_drawer.model_dictionary[self.model_filename]
else: else:
model = load(self.model_path / str(self.model_filename + "_model.joblib")) model = load(self.data_path / str(self.model_filename + "_model.joblib"))
if Path(self.model_path / str(self.model_filename + if Path(self.data_path / str(self.model_filename +
"_svm_model.joblib")).resolve().exists(): "_svm_model.joblib")).resolve().exists():
self.svm_model = load(self.model_path / str(self.model_filename + "_svm_model.joblib")) self.svm_model = load(self.data_path / str(self.model_filename + "_svm_model.joblib"))
assert model, ( assert model, (
f"Unable to load model, ensure model exists at " f"Unable to load model, ensure model exists at "
f"{self.model_path} " f"{self.data_path} "
) )
if self.config["freqai"]["feature_parameters"]["principal_component_analysis"]: if self.config["freqai"]["feature_parameters"]["principal_component_analysis"]:
self.pca = pk.load( self.pca = pk.load(
open(self.model_path / str(self.model_filename + "_pca_object.pkl"), "rb") open(self.data_path / str(self.model_filename + "_pca_object.pkl"), "rb")
) )
return model return model
@ -539,9 +549,9 @@ class FreqaiDataKitchen:
logger.info(f'PCA reduced total features from {n_components} to {n_keep_components}') logger.info(f'PCA reduced total features from {n_components} to {n_keep_components}')
if not self.model_path.is_dir(): if not self.data_path.is_dir():
self.model_path.mkdir(parents=True, exist_ok=True) self.data_path.mkdir(parents=True, exist_ok=True)
pk.dump(pca2, open(self.model_path / str(self.model_filename + "_pca_object.pkl"), "wb")) pk.dump(pca2, open(self.data_path / str(self.model_filename + "_pca_object.pkl"), "wb"))
return None return None
@ -717,40 +727,51 @@ class FreqaiDataKitchen:
return full_timerange return full_timerange
def check_if_new_training_required(self, trained_timerange: TimeRange, def check_if_new_training_required(self, trained_timestamp: int) -> Tuple[bool, TimeRange]:
metadata: dict) -> Tuple[bool, TimeRange]:
time = datetime.datetime.now(tz=datetime.timezone.utc).timestamp() time = datetime.datetime.now(tz=datetime.timezone.utc).timestamp()
trained_timerange = TimeRange()
if trained_timerange.startts != 0: if trained_timestamp != 0:
elapsed_time = (time - trained_timerange.stopts) / SECONDS_IN_DAY elapsed_time = (time - trained_timestamp) / SECONDS_IN_DAY
retrain = elapsed_time > self.freqai_config.get('backtest_period') retrain = elapsed_time > self.freqai_config.get('backtest_period')
if retrain: if retrain:
trained_timerange.startts += self.freqai_config.get( trained_timerange.startts = int(time - self.freqai_config.get(
'backtest_period', 0) * SECONDS_IN_DAY 'backtest_period', 0) * SECONDS_IN_DAY)
trained_timerange.stopts += self.freqai_config.get( trained_timerange.stopts = int(time)
'backtest_period', 0) * SECONDS_IN_DAY
else: # user passed no live_trained_timerange in config else: # user passed no live_trained_timerange in config
trained_timerange = TimeRange()
trained_timerange.startts = int(time - self.freqai_config.get('train_period') * trained_timerange.startts = int(time - self.freqai_config.get('train_period') *
SECONDS_IN_DAY) SECONDS_IN_DAY)
trained_timerange.stopts = int(time) trained_timerange.stopts = int(time)
retrain = True retrain = True
if retrain: # if retrain:
coin, _ = metadata['pair'].split("/") # coin, _ = metadata['pair'].split("/")
# set the new model_path # # set the new data_path
self.model_path = Path(self.full_path / str("sub-train" + "-" + # self.data_path = Path(self.full_path / str("sub-train" + "-" +
str(int(trained_timerange.stopts)))) # str(int(trained_timerange.stopts))))
self.model_filename = "cb_" + coin.lower() + "_" + str(int(trained_timerange.stopts)) # self.model_filename = "cb_" + coin.lower() + "_" + str(int(trained_timerange.stopts))
# this is not persistent at the moment TODO # # this is not persistent at the moment TODO
self.freqai_config['live_trained_timerange'] = str(int(trained_timerange.stopts)) # self.freqai_config['live_trained_timerange'] = str(int(trained_timerange.stopts))
# enables persistence, but not fully implemented into save/load data yer # # enables persistence, but not fully implemented into save/load data yer
self.data['live_trained_timerange'] = str(int(trained_timerange.stopts)) # self.data['live_trained_timerange'] = str(int(trained_timerange.stopts))
return retrain, trained_timerange return retrain, trained_timerange
def set_new_model_names(self, metadata: dict, trained_timerange: TimeRange):
coin, _ = metadata['pair'].split("/")
# set the new data_path
self.data_path = Path(self.full_path / str("sub-train" + "-" +
metadata['pair'].split("/")[0] +
str(int(trained_timerange.stopts))))
self.model_filename = "cb_" + coin.lower() + "_" + str(int(trained_timerange.stopts))
# this is not persistent at the moment TODO
self.freqai_config['live_trained_timerange'] = str(int(trained_timerange.stopts))
# enables persistence, but not fully implemented into save/load data yer
self.data['live_trained_timerange'] = str(int(trained_timerange.stopts))
def download_new_data_for_retraining(self, timerange: TimeRange, metadata: dict) -> None: def download_new_data_for_retraining(self, timerange: TimeRange, metadata: dict) -> None:
exchange = ExchangeResolver.load_exchange(self.config['exchange']['name'], exchange = ExchangeResolver.load_exchange(self.config['exchange']['name'],

102
freqtrade/freqai/freqai_interface.py
View File

@ -13,6 +13,7 @@ from pandas import DataFrame
from freqtrade.configuration import TimeRange from freqtrade.configuration import TimeRange
from freqtrade.enums import RunMode from freqtrade.enums import RunMode
from freqtrade.freqai.data_drawer import FreqaiDataDrawer
from freqtrade.freqai.data_kitchen import FreqaiDataKitchen from freqtrade.freqai.data_kitchen import FreqaiDataKitchen
from freqtrade.strategy.interface import IStrategy from freqtrade.strategy.interface import IStrategy
@ -65,11 +66,14 @@ class IFreqaiModel(ABC):
self.training_on_separate_thread = False self.training_on_separate_thread = False
self.retrain = False self.retrain = False
self.first = True self.first = True
if self.freqai_info.get('live_trained_timerange'): # if self.freqai_info.get('live_trained_timerange'):
self.new_trained_timerange = TimeRange.parse_timerange( # self.new_trained_timerange = TimeRange.parse_timerange(
self.freqai_info['live_trained_timerange']) # self.freqai_info['live_trained_timerange'])
else: # else:
self.new_trained_timerange = TimeRange() # self.new_trained_timerange = TimeRange()
self.set_full_path()
self.data_drawer = FreqaiDataDrawer(Path(self.full_path))
def assert_config(self, config: Dict[str, Any]) -> None: def assert_config(self, config: Dict[str, Any]) -> None:
@ -86,7 +90,7 @@ class IFreqaiModel(ABC):
def start(self, dataframe: DataFrame, metadata: dict, strategy: IStrategy) -> DataFrame: def start(self, dataframe: DataFrame, metadata: dict, strategy: IStrategy) -> DataFrame:
""" """
Entry point to the FreqaiModel, it will train a new model if Entry point to the FreqaiModel from a specific pair, it will train a new model if
necessary before making the prediction. necessary before making the prediction.
The backtesting and training paradigm is a sliding training window The backtesting and training paradigm is a sliding training window
with a following backtest window. Both windows slide according to the with a following backtest window. Both windows slide according to the
@ -103,8 +107,8 @@ class IFreqaiModel(ABC):
self.live = strategy.dp.runmode in (RunMode.DRY_RUN, RunMode.LIVE) self.live = strategy.dp.runmode in (RunMode.DRY_RUN, RunMode.LIVE)
self.pair = metadata["pair"] # FreqaiDataKitchen is reinstantiated for each coin
self.dh = FreqaiDataKitchen(self.config, dataframe, self.live) self.dh = FreqaiDataKitchen(self.config, self.data_drawer, self.live, metadata["pair"])
if self.live: if self.live:
# logger.info('testing live') # logger.info('testing live')
@ -113,7 +117,7 @@ class IFreqaiModel(ABC):
return (self.dh.full_predictions, self.dh.full_do_predict, return (self.dh.full_predictions, self.dh.full_do_predict,
self.dh.full_target_mean, self.dh.full_target_std) self.dh.full_target_mean, self.dh.full_target_std)
logger.info("going to train %s timeranges", len(self.dh.training_timeranges)) logger.info(f'Training {len(self.dh.training_timeranges)} timeranges')
# Loop enforcing the sliding window training/backtesting paradigm # Loop enforcing the sliding window training/backtesting paradigm
# tr_train is the training time range e.g. 1 historical month # tr_train is the training time range e.g. 1 historical month
@ -129,9 +133,12 @@ class IFreqaiModel(ABC):
self.training_timerange = tr_train self.training_timerange = tr_train
dataframe_train = self.dh.slice_dataframe(tr_train, dataframe) dataframe_train = self.dh.slice_dataframe(tr_train, dataframe)
dataframe_backtest = self.dh.slice_dataframe(tr_backtest, dataframe) dataframe_backtest = self.dh.slice_dataframe(tr_backtest, dataframe)
logger.info("training %s for %s", self.pair, tr_train) logger.info("training %s for %s", metadata["pair"], tr_train)
self.dh.model_path = Path(self.dh.full_path / str("sub-train" + "-" + str(tr_train))) trained_timestamp = TimeRange.parse_timerange(tr_train)
if not self.model_exists(self.pair, training_timerange=tr_train): self.dh.data_path = Path(self.dh.full_path /
str("sub-train" + "-" + metadata['pair'].split("/")[0] +
str(int(trained_timestamp.stopts))))
if not self.model_exists(metadata["pair"], trained_timestamp=trained_timestamp.stopts):
self.model = self.train(dataframe_train, metadata) self.model = self.train(dataframe_train, metadata)
self.dh.save_data(self.model) self.dh.save_data(self.model)
else: else:
@ -161,36 +168,40 @@ class IFreqaiModel(ABC):
""" """
self.dh.set_paths() (model_filename,
trained_timestamp,
coin_first) = self.data_drawer.get_pair_dict_info(metadata)
file_exists = self.model_exists(metadata['pair'], if trained_timestamp != 0:
training_timerange=self.freqai_info[ self.dh.set_paths(trained_timestamp)
'live_trained_timerange']) # data_drawer thinks the file eixts, verify here
file_exists = self.model_exists(metadata['pair'],
trained_timestamp=trained_timestamp,
model_filename=model_filename)
if not self.training_on_separate_thread: if not self.training_on_separate_thread:
# this will also prevent other pairs from trying to train simultaneously. # this will also prevent other pairs from trying to train simultaneously.
(self.retrain, (self.retrain,
self.new_trained_timerange) = self.dh.check_if_new_training_required( new_trained_timerange) = self.dh.check_if_new_training_required(
self.new_trained_timerange, trained_timestamp)
metadata) self.dh.set_paths(new_trained_timerange.stopts)
else: else:
logger.info("FreqAI training a new model on background thread.") logger.info("FreqAI training a new model on background thread.")
self.retrain = False self.retrain = False
if self.retrain or not file_exists: if self.retrain or not file_exists:
if self.first: if coin_first:
self.train_model_in_series(self.new_trained_timerange, metadata, strategy) self.train_model_in_series(new_trained_timerange, metadata, strategy)
self.first = False
else: else:
self.training_on_separate_thread = True # acts like a lock self.training_on_separate_thread = True # acts like a lock
self.retrain_model_on_separate_thread(self.new_trained_timerange, self.retrain_model_on_separate_thread(new_trained_timerange,
metadata, strategy) metadata, strategy)
self.model = self.dh.load_data() self.model = self.dh.load_data(coin=metadata['pair'])
strategy_provided_features = self.dh.find_features(dataframe) strategy_provided_features = self.dh.find_features(dataframe)
if strategy_provided_features != self.dh.training_features_list: if strategy_provided_features != self.dh.training_features_list:
self.train_model_in_series(self.new_trained_timerange, metadata, strategy) self.train_model_in_series(new_trained_timerange, metadata, strategy)
preds, do_preds = self.predict(dataframe, metadata) preds, do_preds = self.predict(dataframe, metadata)
self.dh.append_predictions(preds, do_preds, len(dataframe)) self.dh.append_predictions(preds, do_preds, len(dataframe))
@ -252,24 +263,34 @@ class IFreqaiModel(ABC):
if self.freqai_info.get('feature_parameters', {}).get('DI_threshold'): if self.freqai_info.get('feature_parameters', {}).get('DI_threshold'):
self.dh.check_if_pred_in_training_spaces() # sets do_predict self.dh.check_if_pred_in_training_spaces() # sets do_predict
def model_exists(self, pair: str, training_timerange: str) -> bool: def model_exists(self, pair: str, trained_timestamp: int = None,
model_filename: str = '') -> bool:
""" """
Given a pair and path, check if a model already exists Given a pair and path, check if a model already exists
:param pair: pair e.g. BTC/USD :param pair: pair e.g. BTC/USD
:param path: path to model :param path: path to model
""" """
if self.live and training_timerange == "":
return False
coin, _ = pair.split("/") coin, _ = pair.split("/")
self.dh.model_filename = "cb_" + coin.lower() + "_" + training_timerange
path_to_modelfile = Path(self.dh.model_path / str(self.dh.model_filename + "_model.joblib")) if self.live and trained_timestamp is None:
self.dh.model_filename = model_filename
else:
self.dh.model_filename = "cb_" + coin.lower() + "_" + str(trained_timestamp)
path_to_modelfile = Path(self.dh.data_path / str(self.dh.model_filename + "_model.joblib"))
file_exists = path_to_modelfile.is_file() file_exists = path_to_modelfile.is_file()
if file_exists: if file_exists:
logger.info("Found model at %s", self.dh.model_path / self.dh.model_filename) logger.info("Found model at %s", self.dh.data_path / self.dh.model_filename)
else: else:
logger.info("Could not find model at %s", self.dh.model_path / self.dh.model_filename) logger.info("Could not find model at %s", self.dh.data_path / self.dh.model_filename)
return file_exists return file_exists
def set_full_path(self) -> None:
self.full_path = Path(self.config['user_data_dir'] /
"models" /
str(self.freqai_info.get('live_full_backtestrange') +
self.freqai_info.get('identifier')))
@threaded @threaded
def retrain_model_on_separate_thread(self, new_trained_timerange: TimeRange, metadata: dict, def retrain_model_on_separate_thread(self, new_trained_timerange: TimeRange, metadata: dict,
strategy: IStrategy): strategy: IStrategy):
@ -285,7 +306,13 @@ class IFreqaiModel(ABC):
metadata) metadata)
self.model = self.train(unfiltered_dataframe, metadata) self.model = self.train(unfiltered_dataframe, metadata)
self.dh.save_data(self.model)
self.data_drawer.pair_dict[metadata['pair']][
'trained_timestamp'] = new_trained_timerange.stopts
self.dh.set_new_model_names(metadata, new_trained_timerange)
self.dh.save_data(self.model, coin=metadata['pair'])
self.training_on_separate_thread = False self.training_on_separate_thread = False
self.retrain = False self.retrain = False
@ -303,7 +330,14 @@ class IFreqaiModel(ABC):
metadata) metadata)
self.model = self.train(unfiltered_dataframe, metadata) self.model = self.train(unfiltered_dataframe, metadata)
self.dh.save_data(self.model)
self.data_drawer.pair_dict[metadata['pair']][
'trained_timestamp'] = new_trained_timerange.stopts
self.dh.set_new_model_names(metadata, new_trained_timerange)
self.data_drawer.pair_dict[metadata['pair']]['first'] = False
self.dh.save_data(self.model, coin=metadata['pair'])
self.retrain = False self.retrain = False
# Methods which are overridden by user made prediction models. # Methods which are overridden by user made prediction models.

46
freqtrade/freqai/prediction_models/CatboostPredictionModel.py
View File

@ -140,49 +140,3 @@ class CatboostPredictionModel(IFreqaiModel):
# logger.info("--------------------Finished prediction--------------------") # logger.info("--------------------Finished prediction--------------------")
return (self.dh.predictions, self.dh.do_predict) return (self.dh.predictions, self.dh.do_predict)
def data_cleaning_train(self) -> None:
"""
User can add data analysis and cleaning here.
Any function inside this method should drop training data points from the filtered_dataframe
based on user decided logic. See FreqaiDataKitchen::remove_outliers() for an example
of how outlier data points are dropped from the dataframe used for training.
"""
if self.freqai_info.get('feature_parameters', {}).get('principal_component_analysis'):
self.dh.principal_component_analysis()
# if self.feature_parameters["determine_statistical_distributions"]:
# self.dh.determine_statistical_distributions()
# if self.feature_parameters["remove_outliers"]:
# self.dh.remove_outliers(predict=False)
if self.freqai_info.get('feature_parameters', {}).get('use_SVM_to_remove_outliers'):
self.dh.use_SVM_to_remove_outliers(predict=False)
if self.freqai_info.get('feature_parameters', {}).get('DI_threshold'):
self.dh.data["avg_mean_dist"] = self.dh.compute_distances()
def data_cleaning_predict(self, filtered_dataframe: DataFrame) -> None:
"""
User can add data analysis and cleaning here.
These functions each modify self.dh.do_predict, which is a dataframe with equal length
to the number of candles coming from and returning to the strategy. Inside do_predict,
1 allows prediction and < 0 signals to the strategy that the model is not confident in
the prediction.
See FreqaiDataKitchen::remove_outliers() for an example
of how the do_predict vector is modified. do_predict is ultimately passed back to strategy
for buy signals.
"""
if self.freqai_info.get('feature_parameters', {}).get('principal_component_analysis'):
self.dh.pca_transform()
# if self.feature_parameters["determine_statistical_distributions"]:
# self.dh.determine_statistical_distributions()
# if self.feature_parameters["remove_outliers"]:
# self.dh.remove_outliers(predict=True) # creates dropped index
if self.freqai_info.get('feature_parameters', {}).get('use_SVM_to_remove_outliers'):
self.dh.use_SVM_to_remove_outliers(predict=True)
if self.freqai_info.get('feature_parameters', {}).get('DI_threshold'):
self.dh.check_if_pred_in_training_spaces() # sets do_predict