Answer Key

How to use this key

Work the colab through with your partner first, then come here. The code below is one correct answer, not the only one. This was a 45 minute pair exercise with very little scaffolding, so there are several reasonable ways to write almost every cell. If your code looks different but produces the same numbers, you were right.

The written answers matter more than the code. You can already tell whether your code ran. What you cannot check on your own is whether you read the result correctly, and that is what the green Answer boxes are for. Compare your markdown cells against them.

⬅️ Back to the colab

Setup

import pandas as pd

url = 'https://eds-217-essential-python.github.io/data/national_parks.csv'
parks = pd.read_csv(url)
parks.shape

(21560, 12)

parks.columns.tolist()

['year',
 'gnis_id',
 'geometry',
 'metadata',
 'number_of_records',
 'parkname',
 'region',
 'state',
 'unit_code',
 'unit_name',
 'unit_type',
 'visitors']

Part 1: Filtering

1. How many rows describe units whose unit_type is exactly National Park?

national_parks = parks[parks['unit_type'] == 'National Park']
national_parks.shape

(4743, 12)

national_parks['unit_name'].nunique()

61

✅ Answer

4,743 rows, but only 61 National Parks. A row is one park in one year, not one park, and the file covers 113 years of records. If you wrote “there are 4,743 National Parks” you read the row count as a count of things in the world, which it is not.

National Park is also only one of 30 values in unit_type. Most of the file is National Monuments, Historic Sites and the rest of the National Park Service, so this filter throws away about four fifths of the table.

2. How many rows have a year of Total? Use ~ or != to build a table called by_year that has none of them in it. How many rows does by_year have?

totals = parks[parks['year'] == 'Total']
print(totals.shape)

by_year = parks[~(parks['year'] == 'Total')]
print(by_year.shape)

(386, 12)
(21174, 12)

✅ Answer

386 Total rows, leaving 21,174 rows in by_year. The number 386 is not arbitrary: it is exactly the number of distinct units in the file, so every unit carries one and only one Total row.

Those rows are sums of the other rows in the same table. Great Smoky Mountains has a Total of 521,947,058, and adding up its 113 yearly values gives 521,947,058 as well. Leaving them in means counting every visitor twice, once in the year they visited and once in the lifetime sum. Every ranking in Parts 2 and 3 depends on having removed them first.

3. The region column uses two-letter codes. Build a list of the codes PW, IM, and AK, and use .isin() to filter by_year down to units in those three regions. How many rows?

western_regions = ['PW', 'IM', 'AK']
western = by_year[by_year['region'].isin(western_regions)]
western.shape

(9894, 12)

by_year['region'].value_counts()

region
IM    5682
NE    3637
SE    3442
PW    3194
MW    2578
NC    1547
AK    1018
NT      76
Name: count, dtype: int64

✅ Answer

9,894 rows, which is 46.7% of by_year. Three of the eight region codes hold almost half the records, and IM on its own is the largest region at 5,682 rows. That is a fact about where the Park Service has units, not about where people go.

.isin() with a list does the work of three separate comparisons joined by |. The list is the part you edit later, which is why building it as a named variable is worth the extra line.

4. Using two conditions in one filter, find the rows that are National Parks and recorded more than 5 million visitors. How many are there, and what does each row represent?

big = by_year[(by_year['unit_type'] == 'National Park') & (by_year['visitors'] > 5_000_000)]
print(big.shape)
big[['year', 'unit_name', 'state', 'visitors']].sort_values('visitors', ascending=False).head()

(59, 12)

	year	unit_name	state	visitors
794	2016	Great Smoky Mountains National Park	NC	11312786.0
1172	2015	Great Smoky Mountains National Park	NC	10712674.0
6988	1999	Great Smoky Mountains National Park	NC	10283598.0
10952	1987	Great Smoky Mountains National Park	NC	10209841.0
6639	2000	Great Smoky Mountains National Park	NC	10175812.0

big['unit_name'].value_counts()

unit_name
Great Smoky Mountains National Park    55
Grand Canyon National Park              2
Yosemite National Park                  1
Acadia National Park                    1
Name: count, dtype: int64

✅ Answer

59 rows, and each one is a single park in a single year. So the honest sentence is “there were 59 park-years above 5 million visitors”, not “59 parks passed 5 million”.

The distinction matters because only four parks ever did it. Great Smoky Mountains accounts for 55 of the 59 rows, Grand Canyon for two (2015 and 2016), Yosemite for one (2016) and Acadia for one (1989). A row count of 59 sounds like a broad group and is in fact one park repeated.

Acadia’s single appearance is worth a second look. It recorded 5,440,952 visitors in 1989, 4,502,283 in 1988, and then 2,339,591 in 1990. A 57% drop in one year almost certainly reflects a change in how the Park Service counted visitors rather than a real change in visits. Keep that in mind before treating any single year as a fact about the world.

Part 2: Ranking

5. Rank all of parks by visitors, largest first, and look at the top 10. Something is wrong with this answer. Say what, and say why in one sentence.

parks.sort_values('visitors', ascending=False).head(10)[
    ['year', 'unit_name', 'unit_type', 'visitors']
]

	year	unit_name	unit_type	visitors
397	Total	Blue Ridge Parkway	Parkway	871922828.0
674	Total	Golden Gate National Recreation Area	National Recreation Area	611031225.0
412	Total	Great Smoky Mountains National Park	National Park	521947058.0
724	Total	Natchez Trace Parkway	Parkway	443145232.0
653	Total	Lake Mead National Recreation Area	National Recreation Area	411700377.0
507	Total	George Washington Memorial Parkway	National Parkway	330201962.0
690	Total	Gateway National Recreation Area	National Recreation Area	329664174.0
727	Total	Colonial National Historical Park	National Historical Park	282420671.0
599	Total	Lincoln Memorial	National Memorial	238620382.0
676	Total	National Capital Parks	Park	236187845.0

✅ Answer

Every one of the top 10 is a Total row. Sorting the unfiltered table compares 113-year lifetime sums against single years, and a lifetime sum is always the larger number. The largest value, Blue Ridge Parkway at 871,922,828, is not a number of visitors in any year at all.

Notice what the wrong answer looks like. There is no error message. The output is a clean, well-formatted table built from the wrong rows. Nothing in the output tells you the rows are not comparable. You have to know that from having looked at the year column.

6. Fix it, and rank by_year instead. Which single park unit had the highest visitor count in any one year, and in what year?

by_year.sort_values('visitors', ascending=False).head(10)[
    ['year', 'unit_name', 'unit_type', 'state', 'visitors']
]

	year	unit_name	unit_type	state	visitors
11164	1987	Golden Gate National Recreation Area	National Recreation Area	CA	21767176.0
10853	1988	Golden Gate National Recreation Area	National Recreation Area	CA	21759271.0
11475	1986	Golden Gate National Recreation Area	National Recreation Area	CA	21582368.0
5923	2002	Blue Ridge Parkway	Parkway	NC	21538760.0
12942	1981	Golden Gate National Recreation Area	National Recreation Area	CA	20279789.0
10628	1988	Blue Ridge Parkway	Parkway	NC	20002975.0
6274	2001	Blue Ridge Parkway	Parkway	NC	19969587.0
12666	1982	Golden Gate National Recreation Area	National Recreation Area	CA	19897389.0
6973	1999	Blue Ridge Parkway	Parkway	NC	19836842.0
6624	2000	Blue Ridge Parkway	Parkway	NC	19153081.0

✅ Answer

Golden Gate National Recreation Area, in 1987, with 21,767,176 visitors.

The part worth arguing about with your partner is that it is not a National Park. The whole top 10 is Golden Gate National Recreation Area and Blue Ridge Parkway, and no National Park appears anywhere in it. The question asked about park units, and units include parkways, recreation areas and memorials, most of which are free, urban, and easy to drive through.

If you answered “Great Smoky Mountains”, you answered a different and narrower question. That is not a small slip. It is the same ambiguity question 12 asks you to name.

7. Filter to National Parks in the year 2016, then find the top 10 by visitors. Show only the unit_name, state, and visitors columns.

np_2016 = by_year[(by_year['unit_type'] == 'National Park') & (by_year['year'] == '2016')].copy()
print(np_2016.shape)

np_2016.sort_values('visitors', ascending=False).head(10)[['unit_name', 'state', 'visitors']]

(61, 12)

	unit_name	state	visitors
794	Great Smoky Mountains National Park	NC	11312786.0
1033	Grand Canyon National Park	AZ	5969811.0
1087	Yosemite National Park	CA	5028868.0
1025	Rocky Mountain National Park	CO	4517585.0
981	Zion National Park	UT	4295127.0
878	Yellowstone National Park	WY	4257177.0
771	Olympic National Park	WA	3390221.0
773	Acadia National Park	ME	3303393.0
900	Grand Teton National Park	WY	3270076.0
1139	Glacier National Park	MT	2946681.0

✅ Answer

61 National Parks reported a 2016 figure, and Great Smoky Mountains leads with 11,312,786 visitors, close to twice the second-place Grand Canyon at 5,969,811. After that the list descends gently: Yosemite 5,028,868, Rocky Mountain 4,517,585, Zion 4,295,127, and on down to Glacier at 2,946,681.

The shape of the ranking matters more than its order. These ten parks account for 57.5 percent of all 83,920,334 National Park visits in 2016, so the distribution is heavily concentrated at the top and has a long thin tail. Note also that this is a different table from question 6, where no National Park appeared at all.

8. How many of the 61 National Parks recorded more than a million visitors in 2016?

(np_2016['visitors'] > 1_000_000).sum()

np.int64(22)

print(np_2016['visitors'].sum())
print(np_2016[np_2016['visitors'] > 1_000_000]['visitors'].sum() / np_2016['visitors'].sum())
print(np_2016['visitors'].median())

83920334.0
0.8135906132118111
607479.0

✅ Answer

22 of the 61, so just over a third. Those 22 parks account for 81.4% of the 83,920,334 visits recorded to National Parks in 2016, which is the more useful way to say it: a third of the parks receive four fifths of the visits.

The other 39 are much smaller than the top 10 table suggests. The median park recorded 607,479 visitors, and the smallest, Gates of the Arctic in Alaska, recorded 10,047. Any sentence that begins “national parks are crowded” is really a sentence about roughly twenty of them.

Note that (np_2016['visitors'] > 1_000_000).sum() counts True values rather than adding visitors. Summing a boolean Series is the standard way to count matches without building a second table.

9. Use .idxmax() and .loc[] on np_2016 to print the name of the most-visited National Park of 2016 on its own, without displaying the whole row.

busiest = np_2016['visitors'].idxmax()
print(np_2016.loc[busiest, 'unit_name'])
print(np_2016.loc[busiest, 'visitors'])

Great Smoky Mountains National Park
11312786.0

✅ Answer

Great Smoky Mountains National Park, with 11,312,786 visitors, which is 1.9 times Grand Canyon’s 5,969,811 and 13.5% of all National Park visits that year. The gap between first and second place is larger than second place itself.

.idxmax() returned the index label of the largest row, not the value. That is exactly why it is paired with .loc[]: one step answers “which one?” and the next answers “how much?”, and neither step requires you to look at a table with your eyes.

Part 3: Put them together

10. Build a list of the state codes CA, UT, and AZ. Filter np_2016 to National Parks in those three states, rank them by visitors, and display the whole ranked list.

southwest_states = ['CA', 'UT', 'AZ']
southwest = np_2016[np_2016['state'].isin(southwest_states)].copy()

southwest.sort_values('visitors', ascending=False)[['unit_name', 'state', 'visitors']]

	unit_name	state	visitors
1033	Grand Canyon National Park	AZ	5969811.0
1087	Yosemite National Park	CA	5028868.0
981	Zion National Park	UT	4295127.0
1009	Joshua Tree National Park	CA	2505286.0
873	Bryce Canyon National Park	UT	2365110.0
848	Arches National Park	UT	1585718.0
1028	Death Valley National Park	CA	1296283.0
892	Sequoia National Park	CA	1254688.0
1003	Capitol Reef National Park	UT	1064904.0
1100	Saguaro National Park	AZ	820426.0
918	Canyonlands National Park	UT	776218.0
1026	Petrified Forest National Park	AZ	643274.0
973	Kings Canyon National Park	CA	607479.0
804	Redwood National Park	CA	536297.0
1064	Lassen Volcanic National Park	CA	536068.0
917	Channel Islands National Park	CA	364807.0
1023	Pinnacles National Park	CA	215555.0

11. Which of those three states has the park at the top of your list, and which has the most parks in it? Answer both in a markdown cell. You can count the parks per state by eye from your ranked table.

southwest['state'].value_counts()

state
CA    9
UT    5
AZ    3
Name: count, dtype: int64

{state: southwest[southwest['state'] == state]['visitors'].sum() for state in southwest_states}

{'CA': np.float64(12345331.0),
 'UT': np.float64(10087077.0),
 'AZ': np.float64(7433511.0)}

✅ Answer

Arizona has the top park; California has the most parks. Grand Canyon leads the list with 5,969,811 visitors, and Arizona has only three National Parks in total. California has nine, more than Utah’s five and Arizona’s three, and its best park, Yosemite at 5,028,868, is second.

The two questions have different answers, and that is the reason both were asked. Arizona reaches 7,433,511 visits in 2016 with three parks, four fifths of them at one park. California reaches 12,345,331 with nine, of which five recorded under 610,000 apiece. “Which state has the busiest park” and “which state has the most park visiting” and “which state has the most parks” are three different questions, and a single ranked table answers none of them by itself.

These 17 parks are 28% of the 61 National Parks and 35.6% of 2016 National Park visits.

12. In one markdown cell of three or four sentences, answer this: a colleague asks you which national park is “the busiest”. What do you need to know before you can answer, and what would you tell them? Use at least two numbers from today’s work.

y2016 = by_year[by_year['year'] == '2016']
y2016.sort_values('visitors', ascending=False).head(5)[
    ['unit_name', 'unit_type', 'state', 'visitors']
]

	unit_name	unit_type	state	visitors
1055	Golden Gate National Recreation Area	National Recreation Area	CA	15638777.0
779	Blue Ridge Parkway	Parkway	NC	15175578.0
794	Great Smoky Mountains National Park	National Park	NC	11312786.0
889	George Washington Memorial Parkway	National Parkway	VA	10323339.0
1070	Gateway National Recreation Area	National Recreation Area	NY	8651770.0

There is no single right answer here. A strong response names at least two of the three choices that have to be settled before the question means anything, quotes at least two numbers it actually computed, and gives the colleague a usable answer rather than only an objection. Here is one that would earn full marks.

✅ A model answer

Before I can answer I need three things settled. First, which units count: only the 61 National Parks, or all 386 National Park Service units? Second, which year, or whether they mean the all-time total. Third, whether the Total rows have been removed, because leaving them in makes every answer a lifetime sum.

If they mean National Parks in 2016, the answer is Great Smoky Mountains with 11,312,786 visitors, nearly twice Grand Canyon’s 5,969,811. If they mean all Park Service units in 2016, Great Smoky Mountains is third: Golden Gate National Recreation Area recorded 15,638,777 and Blue Ridge Parkway 15,175,578. If they mean the busiest single year ever recorded, it is Golden Gate in 1987 at 21,767,176.

So the answer I would give is: for National Parks, Great Smoky Mountains, and the margin is large. But if they are asking because they want to know where the crowds are, the parkways and recreation areas beat every National Park, and I would want to know what they are planning to do with the number before I picked one.

Where the marks are

If you compare your notebook against this key, look for these four things before you look at anything else.

1. Your by_year has 21,174 rows, and you built it before you ranked anything. Every wrong answer in Part 2 traces back to the 386 Total rows.
2. Your answer to question 6 names a unit that is not a National Park. Golden Gate National Recreation Area in 1987 is the correct answer. If you wrote a National Park, you filtered when the question did not ask you to.
3. You compared year to the string '2016', with quotes. The column is text. Comparing it to the integer 2016 returns zero rows and raises no error, which makes it one of the hardest mistakes in pandas to notice.
4. Your answers say what a row is. 4,743 rows is not 4,743 parks, and 59 rows above 5 million is four parks, not 59. Every count in this file is a count of park-years until you say otherwise.

⬅️ Back to the colab