Neptune-Automate

Description

Simplify your machine learning code with this module

Install

To install the module use:

pip install neptune-automate

Application

Clean your data and handle null values:

(1) Import

import pandas as pd
from automate.DataPreprocessor import Null_Value_Remover

(2) Use the function with your data:

data = pd.read_csv('data.csv')
data_clean = Null_Value_Remover.remove_null_value(data, data['Timestamp_or_Categorical Column'])

Apply different machine learning models:

(1) Import

from automate.machine_learning import Machine_Learning_Models

(2) Use the function with your data:

from automate.machine_learning import Machine_Learning_Models

X = ...
y = ...

Machine_Learning_Models.apply_linear_regression(X, y)
Machine_Learning_Models.apply_random_forest_regression(X, y)
Machine_Learning_Models.apply_decision_tree_regression(X, y)
Machine_Learning_Models.apply_support_vector_regression(X, y)
Machine_Learning_Models.apply_GBR(X, y)
Machine_Learning_Models.apply_LGBM(X, y)
Machine_Learning_Models.apply_XGB(X, y)
Machine_Learning_Models.apply_multinomial_nb(X, y)
Machine_Learning_Models.apply_gaussian_nb(X, y)

Use ARIMA, SARIMA, or SARIMAX for time series analysis:

(1) Import

from automate.statistical_models import Statistics_Models

(2) Use the function with your data:

Statistics_Models.apply_ARIMA(data_clean, p, d, q, 'Target_variable')
Statistics_Models.apply_SARIMA(data_clean, p, d, q, 'Target_variable')
Statistics_Models.apply_SARIMAX(data_clean, p, d, q, 'Target_variable', exog_vars=['variable_1', 'variable_2', 'variable_3'])

Perform exponential smoothing for time series forecasting:

(1) Import

from automate.exponential_smoothening import Exponential_Smoothening

(2) Use the function with your data:

Exponential_Smoothening.simple_exponential_smoothening(data_clean['Target_variable '])
Exponential_Smoothening.holt_smoothening(data_clean['Target_variable'])
Exponential_Smoothening.exponential_smoothening(data_clean['Target_variable'])

Augmented Dickey-Fuller Test:

(1) Import

from automate.augmented_dickey_fuller_test import Augmented_Dickey_Fuller_Test

(2) Use the function with your data:

Augmented_Dickey_Fuller_Test.check_stationarity(data_clean['Target_variable'])

Convert your time series data to stationary:

(1) Import

from automate.augmented_dickey_fuller_test import Stationary_Converter

(2) Use the function with your data:

stationary_data = Stationary_Converter.convert_to_stationary(data_clean['Target_variable'])
Augmented_Dickey_Fuller_Test.check_stationarity(stationary_data)

Evaluation Metrics

The function will allow you to print the evaluation metrics based on the problem dataset.

(1) Import

from automate.evaluation.metrics import print_metrics

(2) Use the function with your data:

# Example for regression task
print_metrics(y_test, y_pred)

# Example for classification task
print_metrics(y_true, y_pred)

Note

This module currently only works for Non-Categorical Time Series Data.

Requirements

pandas
numpy
scikit-learn
statsmodels
xgboost
lightgbm
catboost
matplotlib
seaborn