Types of Trends (Time Series)

Hi Guys! Welcome to the introduction to time series trends. This article is completely for Time Series beginners. If you don't know what is time series, this article is completely for you! After you complete this article, you will be familiar with Time Series fundamentals.

Time Series is an order of data containing time within it. It is all about data visualization and understanding the trends in your data. Time Series comprises date, year, and time. For example, monthly sales data, wind speed data, and heart rate for a person regarding data and time.

The DataSet contains time and date within it, so it is Time Series Dataset. DataSet also has a feature temperature. It is also a time series data because it has time intervals on it.

As I told you, Time Series is all about data visualization, and understanding the trends in your data is very important.

The major assumption is the data that follows the past pattern will follow in the future as well.

Let us try to understand the trends in Time Series Data. Trends are nothing but patterns. The data follows some patterns in your dataset; it means it has some trends within it. Trends have lots of varieties. We will see some of the important trends often used in industries as well.

Positive Secular Trend or Upward Trend:

• It means both the x and y-axis follow the same pattern (Continuously Increasing) or increase trend within it.

• If you see the graph, it has a positive increasing trend within it, so it is a positive secular trend or upward trend.
• What is a positive trend telling us? It tells that healthcare and global equalities are having an increasing trend between Dec-10 to Dec-12. Our assumption is this trend may follow in the future as well so we can predict the future predict the health care trend for the future using some Arima models (algorithm to find future trends).

Negative Secular Trend or Downward Trend:

• It means both the x and y-axis follow the same pattern (continuously decreasing) or decreasing trend within it. Both the positive trend and negative trend are the same, but it has opposite with each other.

• If you see the graph, it has a negative decreasing trend within it, so it is a negative secular trend or downward trend.
• What is a negative trend telling us? It tells that annual mean humidity and the year between 1953 to 2013 has a negative trend. Our assumption is this trend will follow in the future (decreasing trend) so we can predict the future annual mean humidity using some Time Series algorithm.

Variation:

• It is also one type of trend, but it has peaks and troughs. Peak means increasing point and troughs means decreasing point. By the use of peaks and troughs, we can understand where our points will increase and decrease. Another name of peaks and troughs is variability.

• See the picture, you can easily understand what is peaks and troughs.
• In real-world data, contains lots of peaks and troughs in it.
• We will measure various things in our data using peaks and troughs. Make sure you know this concept clearly to understand the future trends. We will see various trends by using these peaks and troughs.

Seasonality:

• Seasonality refers to seasonal data, or it has regular peaks and troughs. Seasonality is not what you think about, like (autumn or spring). It is something like Monday sales of petrol are very high and every winter the sales of warmer is high.

• The basic assumption is the trend followed by the past may or may not follow in the future. By using some Arima models, we can predict the future seasonality trends.
• Seasonality has peaks and troughs, but it is predictable.
• Seasonality is very important. It is present everywhere. Make sure you understand this topic.
• See the image, it has peaks and troughs every year and it is a positive secular trend because it is increasing gradually!

Cyclic:

• Cyclic is also the same as Seasonality, but it is not predictable. It also has peaks and troughs but not in any seasonal trend and it forms for no reason.

• If you see the image, it has many peaks and troughs, but it has no positive trend and no seasonality present in the data.
• Fun Fact: Stock market analysis comes under cyclic trends for this only people cannot predict the stock price.

• See the image.
• Because Stock market analysis is cyclic, we can analyze the data and do some actions but we can't predict future stock prices.
• So Bad right!

Find the trend in this image?

• It has a cyclic trend and no trend within it.
• Try to find more trends on your own. It will help you understand many things in the Time Series concept.

Random:

• It doesn't have any obvious peaks and troughs happening for particular reasons. The peaks and troughs occurred randomly all over the data and it does not have any pattern as well.

• See the image, it has no seasonality within it and the peaks and troughs occur randomly, but it has a negative secular trend.

Small comparison:

Seasonal - It has peaks and troughs (predictable).

Cyclic - It has peaks and troughs (not predictable).

Random - It doesn't have a pattern and it won't have?any peaks and troughs it occurs at random. The variation is just all over the graph.

Find the trend?

• It has a random trend within it.
• NO, specific variation in it.

See some more examples to make sure you know these trends clearly!

Example: 1

Example: 2

This is all about trends in Time Series. Try to practice more trends and visualize the trends and find the variation present in them. It helps to increase data visualization skills.

+

Name: R. Aravindan

Company: Artificial Neurons.AI

Position: Content Writer

Thank you so much!