时序数据分析 — 趋势预测

分析时间序列数据：趋势检测、周期性分析、异常点识别、简单预测。适用于监控数据和业务指标。

Wat ===

Simple library for doing time series analysis in Node.js. Much of the functionality has been closed from the time series module in Ruby's statsample library: https://github.com/clbustos/statsample

Install =======

Easy:

npm install time-series

Usage =====

Simple statistics:

ts = new TimeSeries([1, 2, 3, 4]);

ts.mean(); // => gives 2.5 ts.sd(); // => gives around 1.291 ts.var(); // => gives around 1.667

Moving averages:

ts = new TimeSeries(_.range(30));

// Default MA length is 10, gives 9 null observations at the start ts.ma(); // => [null, ..., null, 4.5, 5.5, 6.5, ..., 23.5, 24.5]

// Different MA length ts.ma(5); // => [null, ..., null, 2, 3, 4, 5, ...]

Exponential moving averages:

ts = new TimeSeries(_.range(30));

ts.ema(); // => [null, ..., null, 5.5, 6.5, 7.5, ...] ts.ema(5); // => [null, ..., null, 3, 4, 5, 6, ...]

Licence =======

MIT

安装

# 安装到当前项目
npx skills add time-series

# 全局安装
npx skills add time-series -g

来源

• 作者：Rob Britton
• 仓库：https://github.com/robbrit/node-time-series
• npm 包：time-series