2
$\begingroup$

not sure if this is the right place or not to ask for advise about my issue, if not sorry you can close this post.

I have a project at university where I have to analyse a dataset with meteorological data and the only requirement is to apply some statistical technique, I don't really have to answer a scientific question.

The data I have are: pressure, air temperature, dew point, relative humidity, wind mean, wind direction, wind gust, illumination parameter, fog parameter and timestamp.

The time range is over a couple of years with a step of 1 minute and the spatial domain is just one site.

The dataset presents a lot of missed data, where the largest gap is of a couple of months.

Based on the dataset, I was thinking, in a first stage, to try to fill the missed data, not with a simple linear regression, but with a more sophisticated algorithm (matrix-based or patter-based algorithm), and then train a machine learning model for fog prediction and compare the results with the fog parameter.

However, atmospheric and meteorological data analysis is not my area of expertise, I don't know if with this data other type of analysis can be done.

Could anyone suggest any other type of analysis can be done with this data?

Improve this question
$\endgroup$
4
  • 1
    $\begingroup$ if you are doing frequency analysis look into lomb scargle analysis for FFT of dataset with missing points $\endgroup$ Commented Apr 28 at 13:08
  • 1
    $\begingroup$ You might consider a year-by-year analysis to identify trends/differences in seasonality. Wind roses are also very informative (e.g. by season). You could also do the fog analysis if that's important at your location. $\endgroup$ Commented Apr 30 at 1:34
  • $\begingroup$ Those are really two great advices, thanks a lot to both of you! $\endgroup$ Commented Apr 30 at 7:59
  • $\begingroup$ @f.thorpe what do you mean by "fog analysis"? the fog prediction? $\endgroup$ Commented Apr 30 at 8:00

1 Answer 1

Reset to default
1
$\begingroup$

Statistical analysis searches for the confidence to report trends or failures to meet trends. Basically, you should ask ... When and how does a change in variable Y correlate with a change in variable X? Now, make variable Y and variable X relatable to real world events. Even as you are saying ... not my area of expertise ... you might already be aware of these "trends" ...

  • Cold + dry, cold + damp, wet + hot, and hot + dry are associations made about weather conditions. How would you establish whether and when these associations are reasonable statements within statistical confidence levels?

  • High pressure fronts bring calm weather. Low pressure fronts bring ... Are these statements true? To what confidence level?

  • Take an umbrella with you, because it certainly "smells" like it is going to rain. How would you prove this advice is statistically valid?

What other weather related insights have you gathered as a layman that could be proven as valid or as gossip by a good go-around with statistical analysis on a large data set?

Improve this answer
$\endgroup$

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.