USEP price prediction
Project description
nextstep
Introduction
Nextstep integrates major popular machine learning algorithms, taking all the hassle data analysts or data scientists could possibly face when deal with multiple matching learning packages. At the same time, it lifts the programming constraints by extracting key parameters into a configuration dictionary, empowering less experienced python users the ability to explore machine learning.
Nextstep was originally developed for a data science challenge which involved price prediction. So it has a dedicated module to obtain data (oil and weather). It evolves into a machine learning prediction toolkit.
Installation
First time installation
pip install nextstep
Upgrade to the latest version
pip install nextstep --upgrade
Quick Tutorial
getData module
1. generate oil prices
from nextstep.getData.oil import *
oil_prices.process()
brent_daily.csv and wti_daily.csv will be generated at the current directory. They contain historical oil price until the most recent day.
2. generate weather data
This function relies on an API key from worldweatheronline. It is free for 60 days as of 27/3/2020. It will generate csv data files in the current directory.
from nextstep.getData.weather import weather
config = {
'frequency' : 1,
'start_date' : '01-Jan-2020',
'end_date' : '31-Jan-2020',
'api_key' : 'your api key here',
'location_list' : ['singapore'],
'location_label' : False
}
data = weather(config).get_weather_data()
model module
Every ML model has a unique config. Please fill in accordingly.
random forest
# examples, please fill in according to your project scope
from nextstep.model.random_forest import random_forest
config = {
'label_column' : 'USEP',
'train_size' : 0.9,
'seed' : 66,
'n_estimators' : 10,
'bootstrap' : True,
'criterion' : 'mse',
'max_features' : 'sqrt'
}
random_forest_shell = random_forest(config)
random_forest_shell.build_model(data) # build model
arima
from nextstep.model.arima import arima
config = {
'lag' : 7,
'differencing' : 0,
'window_size' : 2,
'label_column' : 'USEP',
'train_size' : 0.8,
'seed' : 33
}
arima_shell = arima(config)
data = data[-300:] # get most recent readings
arima_shell.autocorrelation(data) # plot autocorrelation to determine p, lag order
arima_shell.partial_autocorrelation(data) # plot partial autocorrelation to determine q, moving average widow size
arima_shell.build_model(data) # build model
# residual plot to check model performance
arima_shell.residual_plot()
arima_shell.residual_density_plot()
Contribution
Contribution in any kind is welcome.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for nextstep-0.0.12-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 6fe95cb2ec9e1db3e8d41d0e419626bd0aa5fe7cacb43097ce7f3dcce0ac198c |
|
MD5 | 33a50656d7a80d2d98a9b7d0e94899e1 |
|
BLAKE2b-256 | d04dcebbf631756ab5381359aebafe06e96ff9c1d12a8ac09c34cf3934d04457 |