This is a description to provide details about arguments of CryptoPredictions API. Cryptopredictions is an open-source toolbox for pose prediction in Python. It is implemented by Amirhossein Alimohammadi and Ahmad Nosratbakhsh.
CryptoPredictions
├── configs
│ ├── hydra
| ├── data
| └── main.yaml -- main config file for data module (Essentially responsible for creating dataloader)
| ├── model
| ├── pv_lstml.py
│ ├── neareset_neighbor.py
| ├── zero_vel.py
| ├── ...
| ├── optimizer
| ├── adam.yaml -- config file for adam optimizer
| ├── sgd.yaml -- config file for stochastic gradient descent optimizer
| ├── ...
| ├── scheduler
| ├── reduce_lr_on_plateau.yaml -- config file for reducing learning_rate on plateau technique arguments
| ├── step_lr.yaml -- config file for step of scheduler arguments
| ├── ...
| ├── visualize.yaml -- config file for visualizer API arguments
| ├── evaluate.yaml -- config file for evaluate API arguments
| ├── preprocess.yaml -- config file for preprocess API arguments
| ├── train.yaml -- config file for train API arguments
| ├── generate_output.yaml -- config file for generate_output API arguments
| └── metrics.yaml -- config file for metrics
|
└─── logging.conf -- logging configurations
Now we will precisely explain each module.
Directory Location: 'configs/hydra/dataset_loader'
common.yaml:
window_size: Number of previous days to be included in observation
train_start_date: Start time of the train dataset
train_end_date: End time of the train dataset
valid_start_date: Start time of the validation dataset
valid_end_date: End time of the validation dataset
features: Regressors you want to keep from the original dataset(it depends on the dataset)
indicators_names: Name of the indicators you want to use
Bitmex.yaml:
name: Name of the dataloader (should not be changed)
binsize: You could select it from ["1m", "5m", "1h", "1d"] (based on our best knowledge 1h and 1d are the best practice)
batch_size: It doesn not really matter but by increasing it you can improve the fetching time
symbol: For filling the symbol please visit https://github.com/alimohammadiamirhossein/CryptoPredictions/blob/main/README.md#Dataset.
Directory Location: 'configs/hydra/model'
arima.yaml:
type: Name of the model (should not be changed)
order: The (p,d,q) order of the model for the number of AR parameters, differences, and MA parameters to use.
is_regression: If the model is classifier set it to be True otherwise set it to be False.
gru.yaml:
type: Name of the model (should not be changed)
hidden_dim: Hidden dimension of the architecture of the model
epochs: Number of epochs to be trained. (if you use bigger dataset e.g. hourly, it is better to set epochs and hidden_dim bigger)
is_regression: If the model is classifier set it to be True otherwise set it to be False.
lstm.yaml:
type: Name of the model (should not be changed)
hidden_dim: Hidden dimension of the architecture of the model
epochs: Number of epochs to be trained. (if you use bigger dataset e.g. hourly, it is better to set epochs and hidden_dim bigger)
is_regression: If the model is classifier set it to be True otherwise set it to be False.
orbit.yaml:
type: Name of the model (should not be changed)
response_col: Name of the column you want to predict(it depends on the dataset but its default is prediction)
date_col: Name of the date column(it depends on the dataset but its default is Date)
estimator: Name of the estimator {'stan-mcmc', 'stan-map'} – default to be ‘stan-mcmc’
seasonality: Length of seasonality
seed: Random seed
global_trend_option: Transformation function for the shape of the forecasted global trend. { 'linear', 'loglinear', 'logistic', 'flat'}
n_bootstrap_draws: Number of samples to bootstrap in order to generate the prediction interval.
is_regression: If the model is classifier set it to be True otherwise set it to be False.
prophet.yaml:
type: Name of the model (should not be changed)
response_col: Name of the column you want to predict(it depends on the dataset but its default is prediction)
date_col: Name of the date column(it depends on the dataset but its default is Date)
is_regression: If the model is classifier set it to be True otherwise set it to be False.
random_forest.yaml:
type: Name of the model (should not be changed)
n_estimators: The number of trees in the forest.
random_state: Controls both the randomness of the bootstrapping of the samples used when building trees.
is_regression: If the model is classifier set it to be True otherwise set it to be False.
sarimax.yaml:
type: Name of the model (should not be changed)
order: The (p,d,q) order of the model for the number of AR parameters, differences, and MA parameters.
seasonal_order: The (P,D,Q,s) order of the seasonal component of the model for the AR parameters, differences, MA parameters, and periodicity.
enforce_invertibility: Whether or not to transform the MA parameters to enforce invertibility in the moving average component of the model.
enforce_stationarity: Whether or not to transform the AR parameters to enforce stationarity in the autoregressive component of the model.
is_regression: If the model is classifier set it to be True otherwise set it to be False.
xgboost.yaml:
type: Name of the model (should not be changed)
************************************************************
is_regression: If the model is classifier set it to be True otherwise set it to be False.
Check preprocessing config file: "configs/hydra/train.yaml" for more details.
You can change trainer via commandline like below:
model: Name of the model
dataset_loader: Name of the dataset_loader
validation_method: You could select the method of validation from {simple, cross_validation}
load_path: Path to load a dataset
save_dir: Path to save the model
File Location: 'configs/hydra/metrics.yaml'
metrics.yaml:
metrics: You can add metrics to you evaluation.{accuracy_score, f1_score, recall_score, precision_score, rmse}