Time Series Main
Definition:
- Forecasting
- Goals
- Planning
- short/medium/long-term forecasts
Time Series Data EDA
About time series Data:
- A time series is a set of observations
of a time-dependent variable , at a specific time . - It’s most common to consider data collected from discrete time stamps.
- Time Series Analysis to get helpful information from the historic
data
Examine the histrorical data:
- Trend
- the long-term pattern of a time series
- it can be positive or negative
- if there’s no show an increasing or decreasing pattern then the series is stationary
- Seasonality
- A Seasonality occurs when the time series exhibits regular
fluctuations with fixed frequency, like during the same month
- A Seasonality occurs when the time series exhibits regular
- Irregularity:
- An irregular time series is the opposite of a regular time series. The data in the time series follows a temporal sequence, but the measurements might not happen at a regular time interval.
- Cyclic:
- Any pattern showing an up and down movement around a given
trend is identified as a cyclical pattern.
- Any pattern showing an up and down movement around a given
- Anomalies:
- Anomalies are rare items, events, or patterns that significantly
differ from the majority of the data.
- Anomalies are rare items, events, or patterns that significantly
note:
seasonality vs cyclicality: If the fluctuations are not of a fixed frequency then they are cyclic; if the frequency is unchanging and associated with some aspect of the calendar, then the pattern is seasonal.
Analysis Methods:
- Time series plot
- Moving-average smoothing -> for estimate Trend
- Decomposition -> for estimate seasonal
- ACF and PACF plots
- Missing value imputation
- stationary process
- Auto Correlation
note:
MA smoothing is not MA from ARIMA, use the analysis methods to check if the data is stationary or not
Models:
Stationary model
Non-stationary
Trend:
Trend + Seasonality:
- Traditional statistical models
- Newer statistical models
- ML predictive models
- Recursive Neural Nets
- Ensembles models