Assignment codes for Time Series (2020, FGV).
In this repository, there is the beggining of the course. The rest of the course can be found in this repository.
Here, we analysed and modeled two times series. We also tried to predict the second one.
We analysed Covid-19 deaths in Rio de Janeiro, Brazil, between 2020/03/20 and 2020/08/17.
We tried various models to trend:
- Linear
- Polynomial
- Moving average
- LOWESS
We can see, for example, some results from moving average.
We also modeled seasonality:
- Deterministic
- Deterministic and multiplicative
- Stochastic
We modeled car sales from Volvo between 2007 and 2016.
We fitted various models in a window of two years, trying to predict next day. Among them:
- Regression with seasonal dummies
- Polynomial regression
- Exponential smoothing
- Holt
- Additive Holt-Winters
- Multiplicative Holt-Winters
For example, we can see results from Multiplicative Holt-Winters: