This chapter demonstrated a variety of techniques for discovering trends and recognizing anomalies. These two outcomes, trends and anomalies, are related because they both rely on a model that describes the behavior or characteristics of the training data. For discovering trends, we fit a model and then query that model to find data streams that are dramatically increasing or decreasing in recent history. For anomaly detection, we can use the model to forecast the next observation and then check whether the true observation significantly differed from the forecast, or we can query the model to see how normal or abnormal a new observation is compared to the training data. How these techniques are deployed in practice depends on the technique used and the use case, but often one will train a model on recent data (say, the prior 90 days), while taking care to ensure the training data is not tarnished with anomalous data points that can undermine the model's ability to accurately detect trends and anomalies.