xgb_training.py

# developer: Taoshidev
# Copyright © 2023 Taoshi, LLC

# developer: Taoshidev
# Copyright © 2023 Taoshi, LLC

import random

import numpy as np
from sklearn.preprocessing import MinMaxScaler

from mining_objects.xgb_mining_model import BaseMiningModel
from mining_objects.mining_utils import MiningUtils
from time_util.time_util import TimeUtil
from vali_objects.dataclasses.client_request import ClientRequest
from vali_config import ValiConfig

import bittensor as bt

# historical doesnt have timestamps
data_structure = MiningUtils.get_file("/runnable/historical_financial_data/data.pickle", True)
        #data_structure = [data_structure[0][curr_iter:curr_iter+iter_add],
         #                     data_structure[1][curr_iter:curr_iter+iter_add],
          #                    data_structure[2][curr_iter:curr_iter+iter_add],
           #                   data_structure[3][curr_iter:curr_iter+iter_add],
            #                  data_structure[4][curr_iter:curr_iter+iter_add]]
print(len(data_structure[0]))
print("start", TimeUtil.millis_to_timestamp(data_structure[0][0]))
print("end", TimeUtil.millis_to_timestamp(data_structure[0][len(data_structure[0])-1]))
sds_ndarray = np.array(data_structure).T

scaler = MinMaxScaler(feature_range=(0, 1))

scaled_data = scaler.fit_transform(sds_ndarray)
scaled_data = scaled_data.T



# will iterate and prepare the dataset and train the model as provided
prep_dataset = BaseMiningModel.base_model_dataset(scaled_data)
base_mining_model = BaseMiningModel(len(prep_dataset.T)).set_model_dir('./mining_models/xgbTrain.model')
base_mining_model.train(prep_dataset)#, epochs=25)