Autoregression is the use of previous data in order to predict the future data. The prediction is done by taking the weighted sum of the previous values.
Points to remember:
In order to predict values using autoregressive models, the past values should impact future values.
It minimizes the mean squared error.