gs_quant.timeseries.technicals.trend¶
- trend(x, method=SeasonalModel.ADDITIVE, freq=Frequency.YEAR)[source]¶
Trend of series with seasonality and residuals removed.
- Parameters:
x (
Series) – time series with at least two years worth of data.method (
SeasonalModel) – ‘additive’ or ‘multiplicative’. Type of seasonal model to use. ‘multiplicative’ is appropriate when the magnitude of the series’s values affect the magnitude of seasonal swings; ‘additive’ is appropriate when seasonal swings’ sizes are independent of the series’s values.freq (
Frequency) – ‘year’, ‘quarter’, ‘month’, or ‘week’. Period in which full cycle occurs (i.e. the “period” of a wave).
- Return type:
Series- Returns:
date-based time series with trend of input series.
Usage
Uses a centered moving average and convolution to decompose the input series into seasonal, trend, and residual components. This function returns the trend component.
If using the default additive model:
\(Y_t = X_t - S_t - R_t\)
If using the multiplicative model:
\(Y_t = X_t / (S_t * R_t)\)
Examples
Generate price series and compute its trend.
>>> prices = generate_series(1000) >>> trend(prices)
See also