Time Series Prediction
- A time series is a series of points.
- For example, you can have the price of gold, a commodity, or the
number of black bears in Moldova.
- This number changes through time.
- Various techniques can be used to predict how the sequence
will evolve.
- There's an obvious benefit to being able to predict the price
of a stock, so a lot of people are working on it.
- A typical mechanism for solving this problem is linear regression.
- Roughly, given a series of say 10 points each say 1 day apart, draw
a best fit line through them. Extend the line 1 day to project
the value at day 11.
- Linear regression works pretty well in general. It's simple,
fast and easy to understand.
- Any other algorithm really has to do better than this to
be considered a success.
- Note that the particular solution depends on the data.
