Trends

Identifying trends

A trend is a long-term change in data that exists despite the small variations from data point to data point. Although there is no formal limit, if we have a whole year of data, an increase or decrease that lasts for many months could already be considered a trend. It is also possible that there's no trend in the data when the trend line remains horizontal.

One way of identifying a trend is by decomposing a time series into trend, seasonality, and residual (whatever is left after we remove trend and seasonality). This can be done with the seasonal_decompose function from the statsmodels library.