TimesML

This package was developed for time series data analysis and machine learning tasks. The aim of TimesML is to provide high-level APIs that developers and data scientists can easily model their times series data. We plan to support more machine learning models in the future. Thank you for your support, please star⭐ this project if you like.

https://github.com/leodflag/TimesML

https://pypi.org/project/TimesML

pip install TimesML

• Statistics

  This module contains statistics calculation function.

  ex. mean_square_error、coefficient_of_determination、 ACF

• Chart

  Time series data related drawing function.

  ex. statistics_infographic、ACF_chart、forecast_result_group_chart

• Model

  This module contains time series models for forecasting.

  ex. AutoRegressive、SimpleMovingAverage

• ProcessData

  This module contains the access and processing methods of time series data.

  ex. get_data_yahoo、n_order_difference_data、split_data

Please download the file 'g20_new_c.csv' first for the following simple example. https://github.com/leodflag/TimesML/tree/master/test_data

Pay attention to the file path,'g20_new_c.csv' should belong to the same file level as the simple example program.

import TimeSeriesAnalysis.ProcessData as Data
import TimeSeriesAnalysis.Model as Model
import TimeSeriesAnalysis.Chart as Chart
import Math.Statistics as math
# setting parameters
save_path = 'US'
chart = Chart.chart('US')

# read data
data = Data.read_file(path='test_data/g20_new_c.csv', col_name='US')

# contains basic statistics: historocal trend line chart、lag plot、ACF chart. figure 1.
chart.statistics_infographic(data, file_path=save_path, lags=20, xlabel='date', ylabel='population')

# split data into training set and test set
train, test = Data.split_data(data, ratio=0.8)

# autoregressive lag periods :2
model1 = Model.AutoRegressive(lags=2)
model1.fit(train)
model1.predict(test, pure_test_set_predict=True)

# autoregressive lag periods :20
model2 = Model.AutoRegressive(lags=20)
model2.fit(train)
model2.predict(test,pure_test_set_predict= True)

# Save the data predicted by model1 using the test set
Data.save_flie(model1.test_predict, path=save_path, stock_id='US_AR(2)_predict', file_format='csv')

# Combine and compare the prediction results of the two models. figure 2.
chart.forecast_result_group_chart(train, test, model1, model2, file_path=save_path, 
model_1_name='AR(2)', model_2_name='AR(20)', xlabel='date', ylabel='population')

# simple moving average. five days as a unit to calculate the average
model3 = Model.SimpleMovingAverage(windows=5)
model3.fit(data)

# line chart to observe the average situation every five days. figure 3.
chart.line_chart(data, model3.sma_result, chart_title='SMA(5)', file_path=save_path, xlabel='date', ylabel='population')

Describe the diagram drawn by the simple example

historocal trend line chart(Draw historical trend line chart with time series data.)、 lag plot(Scatter plot of lagging periods.)、 ACF chart( Autocorrelation coefficient chart.).

Combine and compare the prediction results of the two models.

line chart to observe the average situation every five days.

https://leodflag.github.io/TimesML_html/

https://colab.research.google.com/drive/1-Y-ZGygL8V0_cgWPISIzP_lBIAwiRu7Y

• Test data set : g20_new_c.csv
  • Dong E, Du H, Gardner L. An interactive web-based dashboard to track COVID-19 in real time. Lancet Inf Dis. 20(5):533-534. doi: 10.1016/S1473-3099(20)30120-1
• Use requests crawl yahoo finance stock price
  • FinLab. 超簡單台股每日爬蟲教學. Retrieved July 22, 2020, from https://www.finlab.tw/超簡單台股每日爬蟲教學