Oracle Interview Questions | Series

Question

You have a series with only one variable “y” measured at time t. How do predict “y” at time t+1? Which approaches would you use?

in progress 0
Dhruv2301 4 years 1 Answer 967 views Great Grand Master 0

Answer ( 1 )

  1. swap007
    0
    Reply

    This is a classic time series problem. Time series generally deals with a value of
    a single variable taken at different time intervals. This time interval can be a day, week, month,
    quarter, year etc. Then, you start by analyzing the time series, trying to understand the trend,
    seasonality, stationarity of the series etc.
    It is necessary to make a time series stationary before you start modeling. Test like ADF(Augmented Dicky-Fuller test)
    can be used to determine stationarity.
    You can achieve stationarity by differencing (taking differences between time intervals), taking logs etc.,
    Once a time series is stationary, you can use models like ARIMA (Auto Regressive Integrated Moving Average)
    to make future predictions.
    ARIMA model takes the form ARIMA(p,d,q) and the values for these parameters can be found out through the
    ACF and PACF plot.

Leave an answer

Browse
Browse