In this article, I use a global warming dataset from Kaggle to demonstrate some of the common time series data preprocessing/analysis methods and time series forecasting models in Python. Similarly to traditional dataset, given a time series dataset, data exploration and preprocessing are required before the time series data can be analyzed, and time series data analysis is required before a time series forecasting model can be chosen and applied to the analyzed dataset for forecasting.

Given a traditional (time order i ndependent) dataset for supervised machine learning for prediction, data exploration and preprocessing are required before features engineering can be performed, and the features engineering needs to be done before a machine learning model can be chosen and applied to the engineered features for prediction. Time series forecasting has many real applications in various areas such as forecasting of business (e.g., sales, stock), weather, decease, and others. Time series include many kinds of real experimental data taken from various domains such as finance, medicine, scientific research (e.g., global warming, speech analysis, earthquakes), etc.

A time series is a sequence of data samples taken in time order with equal time intervals.

Additive sequence of an array