A temporal pattern is a recurring or otherwise meaningful structure associated with time. In time series, it may appear as fluctuations by hour, day, or year, seasonal behavior, a trend, peak activity, or growth at particular locations. section 8.3 section 9.4
Analyze temporal patterns by:
- Checking the time coverage, identifying gaps and inconsistent intervals, and confirming that enough observations exist for comparison or forecasting. section 8.3
- Reshaping measurements to a useful granularity, such as summarizing hourly observations by day, then visualizing the series with line charts. section 9.3.1
- Examining distributions and individual observations to find anomalies before decomposing the series into trend and seasonality. section 9.3.1
- Comparing locations, time periods, or users to identify patterns of interest and reduce analysis to records with sufficiently complete histories. section 8.3
Road traffic and mobile behavior
For bicycle counts, compare morning, weekly, and yearly activity, then look for seasonal patterns and locations with growing traffic. For timestamped mobile events, use browsing times alongside app types and session lengths to build a richer customer segmentation. section 8.3 section 12.3.1
