Time Series Visualizations

Pls Andrew pls pt. 2

What is this?

seasonal plot

When to use a Seasonal Plot

Can be used to visualize seasonality more clearly and have data from each season overlap

What is this?

Seasonal Subseries Plot

When to use a Seasonal Subseries Plot

\-Can be used to study the seasonality

\-Useful if we already know about the period of seasonality

\-Effective way to analyze between and within group seasonal patterns

Characteristics within a time series plot

Seasonality, cycles, trends, and variability

What is this?

connected scatter plot

How to read this connected scatter plot

\-line going from the lower left direction to the upper right indicate positively correlated movement

\-line going from the upper right direction to the lower left indicate negatively (anti) correlated movement

When to use Line graphs

\-Help visualize the temporal trend of the points

\-They are especially useful for visualizing multiple time series on a single plot

\-Lines can be color coded or drawn differently (dotted, solid, dashed) to distinguish the trends from different categories

What is this?

Line Graph

Definition of a line graph

A line graph is formed when we connect the points in a time series plot

What does the x and y axis represent in a time series plot?

In a time series plot, the x-axis represents time and the y-axis represents the variable of interest measured at each time point

What do these dots represent?

These dots represent the number of submissions during the respective month

What is the seasonal pattern of a time series based on the seasonal and seasonal sub-series plot?

When consistent and repeating factors occur within a year (so basically just look at the graph and see if there are repeating patterns)

Seasonality

The repeating patterns that occur within a year; it is a systemic calendar related effect

Cycles

Patterns; when time series data exhibits repeating behavior over an area that is not of fixed frequency (so not seasonal); can only be observed over a long enough period of time

Random Variation (variability)

Noise; random factors that make up a time series

Trend

Exists when there is “movement” (long-term increase or decrease) in the data; there should be an increasing or decreasing slope in your time series

Note: a trend does not have to be linear, it could be quadratic for example

How to read a time series decomposition

When the overall time series is modeled as the sum of each component, it is **additive**. When there is increasing variability, it is **multiplicative**.

Comparing mean vs median filtering for smoothing a time series

Mean is a measure of center that is not robust to outlier and unusual observation. Since the median is robust to outliers, it is a more appropriate filter to use when there are outliers to smooth data

How to interpret the trend-cycle pattern and sources of variability

idk lol

Multiplicative time series

When the original time series is equal to the product of the components

Time series are made up of what two parts?

Signal and Noise