DATA ANALYSIS
Conducting an end-to-end evaluation to reply a socio-economic query with good graphs
Did you are taking a trip round Christmas time? Do you intend to take just a few days off by the tip of March? Or possibly you like taking the chance to journey the world throughout one other time of the yr like summer season?
It doesn’t matter what kind of vacation season traveler you might be, chances are high that you’ll take some holidays this yr. Now that the worst (fingers crossed) of the COVID-19 disaster is over, have you ever puzzled if tourism is again to its pre-crisis stage?
This can be a query I requested myself just lately and I wish to reply it. I discover that conducting analyses on subjects out of the strictly skilled world is an efficient train to remain “within the recreation of knowledge analytics”. Whether or not you’re a junior or a extra superior knowledge analyst, I discover it all the time helpful to coach with new datasets exterior of your each day job.
Let me take you on my journey to conducting an end-to-end evaluation from the primary thought till the ultimate output!
On this article I wish to take the chance to replicate on this yr of journey and to research if tourism is again to its pre-crisis ranges. This might be seen worldwide or the scope might be narrowed right down to a restricted geographical space. As I dwell in Europe I’m notably occupied with analyzing the influence of the COVID disaster on tourism in Europe.
One other axis of my considering is: how do I measure “tourism”? This notion encapsulate varied fields reminiscent of transportation (by aircraft, by prepare, by automotive…), touristic websites (museums, occasions…), lodging (resorts, tenting websites, home-stays…). To be extra exact within the challenge I wish to analyze right here, let’s concentrate on lodging.
Within the current case I don’t have firm knowledge or a predefined dataset accessible, so let’s browse the Web to search out some open knowledge associated to my matter.
Eurostat supplies open datasets that may be visualized on-line and downloaded in a number of codecs. Knowledge from Eurostat can be utilized without spending a dime so long as it’s talked about that Eurostat is the information supply. Right here I’ll use this dataset concerning the nights spent at vacationer lodging institutions: TOUR_OCC_NIM whose supply is Eurostat.
To reply my preliminary query I wish to examine the worldwide evolution of tourism in Europe with the evolution in every nation. By doing so I ought to have the ability to see if, when and by which nation tourism got here again to its pre-crisis stage.
To take action I’ll want two graphs: one displaying the entire variety of nights spent at vacationer lodging institutions per 30 days, the second displaying the identical metric break up by nation.
With my uncooked dataset at hand (step 2) and my goal output in thoughts (step 3), I’m now all set to conduct the evaluation. Did you discover that really getting into in “knowledge evaluation mode” comes at step 4 and never earlier? It is because analyzing knowledge is far more about why you’ll want to do this evaluation than truly doing it.
To attract the primary graph I need to group the values (variety of nights spent at vacationer lodging institutions) by month. For the second graph I need to group by month and by nation, as follows:
SELECT TIME_PERIOD AS month,geo AS nation,SUM(OBS_VALUE) AS nb_nights_spent,FROM my_dataset.raw_dataGROUP BY 1,2ORDER BY 1,2
The output of code snippets are knowledge tables, not (but) graphs. To show these tabular outputs into good charts, let’s use a knowledge visualization instrument. Right here I exploit the mix of Google BigQuery for step 4 and Looker Studio (beforehand referred to as DataStudio) for step 5.
As we beforehand drew the goal output, we already know the way our last graphs ought to seem like. This protects loads of time right here as I solely must configure the instrument to place the suitable dimensions on the proper place. This could give me these graphs:
So what now? Constructing graphs is nice, however with no human mind to interpret the ensuing outputs it’s quite ineffective. Let’s return to our preliminary query: is tourism in Europe again to its pre-crisis stage primarily based on lodging knowledge? We wish to present a solution as clear as potential to this query.
If we take a look at the worldwide evolution tourism in Europe appears to be again on observe to succeed in its pre-crisis stage. Though the 2022 summer season seasons has not precisely attain the values of 2019 (-15% in July-August 2022 versus July-August 2019), the evolution is displaying a constructive pattern in comparison with 2020 and 2021. It might be fascinating to conduct the identical evaluation subsequent yr when the 2023 summer season season is over.
If now we have a better take a look at the evolution of every nation, this normal remark doesn’t all the time apply. For instance Spain values in 2022 are very near 2019 values (-4% just for July-August) whereas for others 2022 is much beneath 2019 (-26% in Czechia).
One other vital ingredient to keep in mind when deciphering outcomes considerations biases. First there might be biases in the best way knowledge is collected: as we examine totally different nations, every nation could apply a distinct methodology to account for the variety of nights spent at vacationer lodging institutions.
Second concluding that tourism is nearly again to its pre-crisis stage solely primarily based on the evaluation of 1 metric is essentially partial. To have the ability to draw a totally correct conclusion concerning the state of tourism in Europe, one ought to analyze a number of metrics, examine the outcomes of those analyses and consolidate the teachings discovered from them.
Briefly: when conducting an evaluation take these 6 steps and watch out for biases.
>> If you wish to get a visible recap of this text, you may obtain it right here FOR FREE <<
