The goal of this project is to find out the correlation between the sunspots and solarflares. My hipothesis is that correlation exists.
The flares dataset is from Kaggle, thanks to Arxiv. The dataset about sunspots is from the Quandl API.
For this project I build a pipeline with all functions to make the statatistics, plots and the report. Just executing the file main.py or main.ipynb all work is done.
The distribution of solar flares in time is shown in the next plot:
The distribution of sunspots is:
The data for this project is from just one solar cycle, with its maximum in 2013, so is needed more info. It is not clear, but that plots look like a Normal Distribution in time. The proportion of both flares and spots are as follow:
Both distributions look like a Poisson Distribution. That makes sense, after all both phenomena are "rare" events. The next plot shows the flares classification. There is no data about A-class flares:
With this analysis, I've tried to find the correlation between solar flares and sunspots. For that I use Pearson, Spearman and Kendall correlations to look for linear and non-linear correlation. First, Pearson correlation:
Pearson correlation between the flares number and spot number is 0.32
Pearson correlation between both monthly averages is 0.5
Pearson correlation between spot monthly average and flares monthly maximum is 0.35
It seems there is no linear correlation between both variables, except in the monthly averages (reference).
Spearman correlation between the flares number and spot number is 0.38
Spearman correlation between both monthly averages is 0.75
Spearman correlation between spot monthly average and flares monthly maximum is 0.71
In this case there is a strong non-linear correlation between the monthly averages and the spot monthly average with flares monthly maximum. Now Kendall correlation:
Kendall correlation between the flares number and spot number is 0.27
Kendall correlation between both monthly averages is 0.56
Kendall correlation between spot monthly average and flares monthly maximum is 0.52
With all this data, effectively there is some kind of correlation between solar flares and sunspots, and therefore between both and coronal mass ejections (CME), which can be cause of serious damage in technological fields. For causality research would be neccesary more analysis with mora data, as the sunspots size or bigger temporal range.