A missing value is an absent observation in a dataset. It matters because calculations can quietly become misleading when the data for a period or category is incomplete. Before combining values by year, check whether every country and year needed by the comparison is represented. section 2.2.1
Incomplete boundary years
Tourist-arrival series contain gaps for some countries and years. The 1990 starting year is unavailable for some countries, while 2019 ends in October rather than covering a full year. Summing or comparing these incomplete years with complete annual totals could distort the apparent trend. section 2.2.1 section 2.3.1
There is no single automatic response to missing values: choose a treatment that fits the analysis goal and the pattern in the data. For a progressively growing time series, excluding incomplete boundary years can provide a comparable range. Here, restricting the analysis to 1994–2018 preserves years with usable data before annual totals are calculated. section 2.3.1
A practical workflow for this time-series case is:
- Load the data into a pandas DataFrame and parse the Date field as dates. section 2.2.1
- Extract the year from Date, then retain only rows from 1994 through 2018. section 2.3.1
- Group retained rows by Year and Country, then sum tourist arrivals and reset the index. section 2.3.1
- Plot the resulting yearly values as Altair line charts, using only the comparable range. section 2.3.1
Table 2.1 Arrivals at tourist arrivals establishments since 1994
|
Date
|
IT
|
FR
|
DE
|
PT
|
ES
|
UK
|
|---|---|---|---|---|---|---|
|
1990-01-01
|
2,543,920
|
3,185,877
|
325,138
|
1723,786
|
1,776,000
|
|
|
1990-02-01
|
2,871,632
|
3,588,879
|
381,539
|
1,885,718
|
2,250,000
|
|
|
1990-03-01
|
3,774,702
|
4,272,437
|
493,957
|
2,337,847
|
2,662,000
|
|
|
…
|
…
|
…
|
…
|
…
|
…
|
…
|
|
2019-08-01
|
11,649,500
|
13,692,822
|
14,570,339
|
2,531,809
|
12,893,366
|
8,889,049
|
|
2019-09-01
|
9,888,817
|
11,684,845
|
14,373,815
|
2,263,748
|
5,858,984
|
|
|
2019-10-01
|
7,692,388
|
10,401,793
|
13,780,441
|
1,995,942
|
7,455,781
|
Inspect the country-by-date table to see why completeness must be checked before yearly aggregation or comparison. section 2.2.1
Filtering incomplete years with pandas
Create a Year column from Date, then filter out years before 1994 and after 2018. This approach removes incomplete endpoints rather than treating partial annual data as directly comparable totals. section 2.3.1
Listing 2.8 Removing missing years from the dataset
df.loc[:, 'Year'] = df['Date'].dt.year #1 df = df[(df['Year'] >= 1994) & (df['Year'] <= 2018)] #2
The filtering listing shows the date-to-year conversion and the inclusive 1994–2018 range restriction. section 2.3.1
Filtering is a decision about what claim a chart can support. A chart of annual growth should compare like with like: complete annual periods. Once the comparable range is selected, yearly sums and percentage increases can support a clearer trend-focused visualization. section 2.3.1
