I am very new to data science, my question is what model should I use to predict for, let's say, the next 2 years if I only have data available for less than a year?

From what I read online, most models require to lag the data for the same time frame as I want to predict it, which I cannot do in this case, any suggestions?

  • 1
    $\begingroup$ There are a million ways to answer this, none of them definitively correct. I suspect you haven't taken a course in linear models. I would try taking that and a course or two in time series analysis plus thinking about the specific problem you are trying to analyze. Right now, your question is so general it is almost impossible to answer. $\endgroup$
    – kurtosis
    Aug 5 '20 at 22:05
  • 1
    $\begingroup$ Here's the best answer: don't. You are extrapolating WAY beyond what would be comfortable. Particularly if your data has any seasonality it in, less than one year's worth of data won't capture that. $\endgroup$ Aug 5 '20 at 22:08
  • $\begingroup$ How many data points do you have? less than a year of monthly data is not the same as less than a year of hourly data $\endgroup$
    – Skander H.
    Aug 5 '20 at 22:39
  • $\begingroup$ I have daily data, for 219 days, some days are skipped, and i am missing data for the last two months to complete a year, so basically i have data for september '19 till july'20. $\endgroup$
    – user293291
    Aug 5 '20 at 23:23
  • 1
    $\begingroup$ A rule of thumb for time series is that you need 50 periods as a minimum generally these are months and years. With seasonality in your data multiple years are particularly important. $\endgroup$
    – user54285
    Aug 6 '20 at 21:32