In recent years, it has been said that climate change is progressing due to the effects of global warming. In this presentation, we will introduce an attempt to detect climate change in various regions of Japan using JMP's process control method.
The weather data collected for analysis includes temperature (average, maximum, minimum) and precipitation (total amount, maximum amount per hour) observed at approximately 60 weather stations across Japan over the past 30 years.
There are various methods for examining fluctuations in time-series data, but in JMP, process control techniques such as control charts can be used to detect process shifts (sudden changes) and drifts (tendencies of slow changes). Therefore, we consider weather data to be a type of "process" and attempt to detect fluctuations.
First, using JMP's "Process Screening," we extract observation locations and observation items (temperature, precipitation) that show shifts or drifts from the process summary table and graph obtained.
For the extracted points and items, we will use control charts and their tests to closely examine whether there have been any recent fluctuations in those items, while keeping in mind the assumptions of the control chart (normality, independence). We will use not only the commonly used Shewhart control chart, but also the CUSUM (cumulative sum) control chart, which is used to detect small shifts in the process, to consider the situation.
