Skip to content

Commit

Permalink
docs: Review docstrings (#199)
Browse files Browse the repository at this point in the history 
* Review docstrings

* Review doscstring

* Arrangement of number of characters in the lines

* Update worldbank.py
  • Loading branch information
@YurelyCamacho @fccoelho
YurelyCamacho authored and fccoelho committed Feb 14, 2023
1 parent cca7894 commit 9fbf73a
Show file tree
Hide file tree
Showing 6 changed files with 273 additions and 213 deletions.
29 changes: 17 additions & 12 deletions epigraphhub/analysis/forecast_models/metrics.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -9,17 +9,21 @@
# computing some metrics
def compute_metrics(df_pred: pd.DataFrame) -> pd.DataFrame:
"""
This function evaluates the predictions obtained in the `train_eval` method
in the train and test sample. The predictions must be saved in a dataset with the following columns:
'median', 'target' and 'train_size'.
This function uses the following metrics:
- explained variance score;
- mean absolute error;
- mean squared error;
- root mean squared error;
- mean squared log error;
- mean absolute percentage error.
To compute this metrics we use the implementations of the sklearn.metrics package.
This function evaluates the predictions obtained in the `train_eval`
method in the train and test sample. The predictions must be saved
in a dataset with the following columns: 'median', 'target' and
'train_size'.
This function uses the following metrics:
- explained variance score;
- mean absolute error;
- mean squared error;
- root mean squared error;
- mean squared log error;
- mean absolute percentage error.
To compute this metrics we use the implementations of the
sklearn.metrics package.
Parameters
----------
Expand All @@ -29,7 +33,8 @@ def compute_metrics(df_pred: pd.DataFrame) -> pd.DataFrame:
Returns
-------
pd.DataFrame
Dataframe with two columns: out_sample and in_sample and with the metrics as index.
DataFrame with two columns: out_sample and in_sample and with
the metrics as index.
"""

metrics = [
Expand Down
141 changes: 78 additions & 63 deletions epigraphhub/analysis/forecast_models/ngboost_models.py
View file
Original file line number Diff line number Diff line change
@@ -1,9 +1,9 @@
#!/usr/bin/env python3
"""
The functions in this module allow the application of the
ngboost regressor model. There are separate methods to train and evaluate (separate
the data in train and test datasets), train with all the data available, and make
forecasts.
The functions in this module allow the application of the ngboost
regressor model. There are separate methods to train and evaluate
(separate the data in train and test datasets), train with all the data
available, and make forecasts.
"""

from typing import Union
Expand Down Expand Up @@ -40,9 +40,9 @@

class NGBModel:
"""
This class allows the user to create a ngboost model. The methods in this
class allows the user to train and evaluate the model, to train and save the model
and make the forecast using saved models.
This class allows the user to create a ngboost model. The methods
allows to train and evaluate the model, to train and save the model
and make the forecast using saved models.
"""

def __init__(
Expand All @@ -57,19 +57,21 @@ def __init__(
Parameters
----------
look_back : int
Number of the last days that will be used to forecast the next days.
Number of the last days that will be used to forecast the
next days.
predict_n : int
Number of days that will be predicted.
validation_split : float
Proportion of training data to set aside as validation data for early stopping.
Proportion of training data to set aside as validation data
for early stopping.
early_stop : int
The number of consecutive boosting iterations during which the
loss has to increase before the algorithm stops early.
Set to None to disable early stopping and validation.
None enables running over the full data set.
params_model : dictionary
The number of consecutive boosting iterations during which
the loss has to increase before the algorithm stops early.
Set to None to disable early stopping and validation. None
enables running over the full data set.
params_model : dict
The dict with the params that will be used in the ngboost
regressor model.
regressor model.
"""

# This will remove the validation_fraction and early_stopping_rounds parameters since it shouldn't.
Expand Down Expand Up @@ -98,44 +100,51 @@ def train_eval(
save: bool = False,
) -> pd.DataFrame:
"""
Function to apply a ngboost regressor model given a dataset and a target column.
This function will train multiple models, each one specilist in predict the X + n
days, of the target column, where n is in the range (1, number of days that you
want predict).
This function split the data in train and test dataset and returns the predictions
made using the test dataset.
Function to apply a ngboost regressor model given a dataset and
a target column. This function will train multiple models, each
one specilist in predict the X + n days, of the target column,
where n is in the range (1, number of days that you want
predict). This function split the data in train and test dataset
and returns the predictions made using the test dataset.
Parameters
----------
target_name : str
Name of the target column.
Name of the target column.
data : pd.DataFrame
Dataframe with features and target column.
DataFrame with features and target column.
ini_date : str, optional
Determines the beggining of the train dataset, by default None
Determines the beggining of the train dataset, by default
None.
end_train_date : str, optional
Determines the beggining of end of train dataset. If end_train_date
is not None, then ratio isn't used, by default None
Determines the beggining of end of train dataset. If is not
None, then ratio isn't used, by default None.
end_date : str, optional
Determines the end of the dataset used in validation, by default None
Determines the end of the dataset used in validation, by
default None.
ratio : float
Determines which percentage of the data will be used to train the model, by default 0.75
Determines which percentage of the data will be used to
train the model, by default 0.75.
path : str, optional
It indicates where save the models trained, by default None
It indicates where save the models trained, by default None.
name : str, optional
It indicates which name use to save the models trained, by default None
It indicates which name use to save the models trained, by
default None.
save : bool
If True the models trained are saved, by default False
If True the models trained are saved, by default False.
Returns
-------
pd.DataFrame
A dataframe with four columns (and a date index):
- target: The target values
- lower: The lower value of the confidence interval of 95%
- median: The median value of the confidence interval of 95%
- upper: The upper value of the confidence interval of 95%
- train_size: The number of rows of data using as training data.
A DataFrame with four columns (and a date index):
- target: The target values.
- lower: The lower value of the confidence interval of 95%.
- median: The median value of the confidence interval of
95%.
- upper: The upper value of the confidence interval of 95%.
- train_size: The number of rows of data using as training
data.
"""

df_lag = preprocess_data(data, self.look_back, ini_date, end_date)
Expand Down Expand Up @@ -251,32 +260,35 @@ def train(
name: str = "train_ngb",
) -> list:
"""
Function to train multiple ngboost regressor models given a dataset and a target column.
This function will train multiple models, each one specilist in predict the X + n
days, of the target column, where n is in the range (1, number of days that you
want predict).
This function will train the model with all the data available and will save the model
that will be used to make forecasts.
Function to train multiple ngboost regressor models given a
dataset and a target column. This function will train multiple
models, each one specilist in predict the X + n days, of the
target column, where n is in the range (1, number of days that
you want predict). This function will train the model with all
the data available and will save the model that will be used to
make forecasts.
Parameters
----------
target_name : str
Name of the target column.
data : pd.DataFrame
Dataframe with features and target column
DataFrame with features and target column.
ini_date : str, optional
Determines the beggining of the train dataset, by default None
Determines the beggining of the train dataset, by default
None.
end_date : str, optional
Determines the end of the train dataset, by default None
Determines the end of the train dataset, by default None.
save : bool
If True the models is saved, by default True
If True the models is saved, by default True.
path : str, optional
Indicates where the models will be saved, by default "../opt/models/saved_models/ml"
Indicates where the models will be saved, by default
"../opt/models/saved_models/ml".
Returns
-------
list
A list with the trained models
A list with the trained models.
"""

predict_n = self.predict_n
Expand Down Expand Up @@ -320,34 +332,37 @@ def forecast(
) -> pd.DataFrame:

"""
Function to load multiple ngboost regressor model trained with the function
`training_model` and make the forecast.
Function to load multiple ngboost regressor model trained with
the function `training_model` and make the forecast.
Important:
predict_n and max_lag need have the same value used in training_model
Only the last that of the dataset will be used to forecast the next
predict_n days.
Important: predict_n and max_lag need have the same value used
in training_model. Only the last that of the dataset will be
used to forecast the next predict_n days.
Parameters
----------
target_name : str
Name of the target column.
data : pd.DataFrame
Dataframe with features and target column
DataFrame with features and target column.
ini_date : str, optional
Determines the beggining of the train dataset, by default None
Determines the beggining of the train dataset, by default
None.
end_date : str, optional
Determines the end of the train dataset, by default None
Determines the end of the train dataset, by default None.
path : str, optional
Indicates where the models will be saved, by default "../opt/models/saved_models/ml"
Indicates where the models will be saved, by default
"../opt/models/saved_models/ml".
Returns
-------
pd.DataFrame
A dataframe with three columns regarding(and a date index):
- lower: The lower value of the confidence interval of 95%
- median: The median value of the confidence interval of 95%
- upper: The upper value of the confidence interval of 95%
A DataFrame with three columns regarding(and a date index):
- lower: The lower value of the confidence interval of 95%.
- median: The median value of the confidence interval of
95%.
- upper: The upper value of the confidence interval of 95%.
"""

df_lag = preprocess_data(data, self.look_back, None, end_date)
Expand Down
Loading

0 comments on commit 9fbf73a

Please sign in to comment.