/Time-Series-Forecasting

Time-Series-Forecasting

Basic Terminologies in Time Series Forecasting

  1. Stationarity
    • It refers to the data that exhibits a consistent statistical distribution over time.
  2. Differencing
    • This is a technique used to make a non-stationary data stationary.
    • Involves taking the difference between consecutive observations to remove the trend or seasonality.
  3. Time interval
    • The fixed amount of time between each observation in a time series. It can be same for the inputs and ouputs or a specific time interval for inputs and different time interval of ouput.
    • I have handled data where Inputs were taken between 1 hour and the ouputs were found between 15mins so that there are 4 outputs in just 1 hour.
  4. Timestamp
    • The specific time at which an observation is recorded.
    • This is usually converted into sine and cosine encoding to make the model understand that this is variation of time.
  5. Seasonality
    • A regular pattern of changes in a time series that occurs at fixed intervals of time.
  6. Autocorrelation
    • The degree to which a time series is correlated with its own lagged values.
  7. Lag
    • The amount of time between two observations in a time series.
  8. Smoothing
    • The process of removing noise and other short term fluctuations from a time series to reveal its underlying trends and seasonality.
  9. Trend
    • The long term pattern or direction in a time series.

References: