Answer: Line chart

Here, there are two quantitative variables (number of points and game number), and one of them involves some element of time (game number). Line charts are appropriate when one quantitative variable is time.

Difficulty: ⭐️⭐️

The average score on this problem was 75%.

Answer:

1. plum.get('Date').str.contains('-06-')

2. get('PTS').sum()

To find the total number of points Kelsey Plum scored in June, one approach is to first create a DataFrame with only the rows for June. During the month of June, the 'Date' values contain '-06-' (since June is the 6th month), so plum.get('Date').str.contains('-06-') is a Series containing True only for the June rows and june_only = plum[plum.get('Date').str.contains('-06-')] is a DataFrame containing only the June rows.

Then, all we need is the sum of the 'PTS' column, which is given by june_only.get('PTS').sum().

Difficulty: ⭐️

The average score on this problem was 90%.

Answer: The number 3

plum.groupby('Opp').max() finds the largest value in the 'Date', 'Home', 'Won', 'PTS', 'AST', and 'TOV' columns for each unique 'Opp' (independently for each column). grouped = plum.groupby('Opp').max().get(['Date', 'PTS']) keeps only the 'Date' and 'PTS' columns. Note that in grouped, the index is 'Opp', the column we grouped on.

When grouped.reset_index() is called, the index is switched back to the default of 0, 1, 2, 3, 4, and so on. Then, grouped.reset_index().index is an Index containing the numbers [0, 1, 2, 3, 4, ...], and np.array(grouped.reset_index().index) is np.array([0, 1, 2, 3, 4, ...]). In this array, the number at position i is just i, so the number at position df is df. Here, df is the argument to unknown, and we were asked for the value of unknown(3), so the correct answer is the number at position 3 in np.array([0, 1, 2, 3, 4, ...]) which is 3.

Note that if we asked for unknown(50) (or unknown(k), where k is any integer above 30), the answer would be “An error”, since grouped could not have had 51 rows. plum has 31 rows, so grouped has at most 31 rows (but likely less, since Kelsey Plum’s team likely played the same opponent multiple times).

Difficulty: ⭐️⭐️⭐️

The average score on this problem was 72%.

Answer: 4 rows and 5 columns.

plum.groupby(['Home', 'Won']).mean() contains one row for every unique combination of 'Home' and 'Won'. There are two values of 'Home' - True and False – and two values of 'Won' – True and False – leading to 4 combinations. We can assume that there was at least one row in plum for each of these 4 combinations due to the assumption given in the problem:

Suppose that Plum’s team, the Las Vegas Aces, won at least one game in Las Vegas and lost at least one game in Las Vegas. Also, suppose they won at least one game in an opponent’s arena and lost at least one game in an opponent’s arena.

plum started with 7 columns: 'Date', 'Opp', 'Home', 'Won', 'PTS', 'AST', and 'TOV'. After grouping by ['Home', 'Won'] and using .mean(), 'Home' and 'Won' become the index. The resulting DataFrame contains all of the columns that the .mean() aggregation method can work on. We cannot take the mean of 'Date' and 'Opp', because those columns are strings, so plum.groupby(['Home', 'Won']).mean() contains a MultiIndex with 2 “columns” – 'Home' and 'Won' – and 3 regular columns – 'PTS' 'AST', and 'TOV'. Then, when using .reset_index(), 'Home' and 'Won' are restored as regular columns, meaning that plum.groupby(['Home', 'Won']).mean().reset_index() has 2 + 3 = 5 columns.

Difficulty: ⭐️⭐️

The average score on this problem was 78%.

Answer: home_won and puzzle have the same number of rows, but a different number of columns

All that changed between home_won and puzzle is the aggregation method. The aggregation method has no influence on the number of rows in the output DataFrame, as there is still one row for each of the 4 unique combinations of 'Home' and 'Won'.

However, puzzle has 7 columns, instead of 5. In the solution to the above subpart, we noticed that we could not use .mean() on the 'Date' and 'Opp' columns, since they contained strings. However, we can use .count() (since .count() just determines the number of non-NA values in each group), and so the 'Date' and 'Opp' columns are not “lost” when aggregating. Hence, puzzle has 2 more columns than home_won.

Difficulty: ⭐️⭐️

The average score on this problem was 85%.

Answer:

1. 'Opp'

2. sort_values('Won').index[0]

The key insight here is that the values in the 'Won' column are Boolean, and when Boolean values are used in arithmetic they are treated as 1s (True) and 0s (False). The sum of several 'Won' values is the same as the number of wins.

If we group plum by 'Opp' and use .sum(), the resulting 'Won' column contains the number of wins that Plum’s team had against each unique opponent. If we sort this DataFrame by 'Won' in increasing order (which is the default behavior of sort_values), the row at the top will correspond to the 'Opp' that Plum’s team had no wins against. Since we grouped by 'Opp', team names are stored in the index, so .index[0] will give us the name of the desired team.

Difficulty: ⭐️⭐️⭐️

The average score on this problem was 67%.

Answer: -(s == False).sum()

n_wins equals the number of wins that Plum’s team had. Recall that her team played 31 games in total. In order for (31 + modify_series(plum.get('Won'))) to be equal to her team’s number of wins, modify_series(plum.get('Won')) must be equal to her team’s number of losses, multiplied by -1.

To see this algebraically, let modified = modify_series(plum.get('Won')). Then:

31 + \text{modified} = \text{wins} \text{modified} = \text{wins} - 31 = -(31 - \text{wins}) = -(\text{losses})

The function modified_series(s) takes in a Series containing the wins and losses for each of Plum’s team’s games and needs to return the number of losses multiplied by -1. s.sum() returns the number of wins, and (s == False).sum() returns the number of losses. Then, -(s == False).sum() returns the number of losses multiplied by -1, as desired.

Difficulty: ⭐️⭐️

The average score on this problem was 76%.

Answer: 0.2

Recall, the TVD is the sum of the absolute differences in proportions, divided by 2. The absolute differences in proportions for each category are as follows:

• Free Throws: |0.05 - 0.2| = 0.15
• Threes: |0.35 - 0.2| = 0.15
• Midrange: |0.35 - 0.3| = 0.05
• Layups: |0.25 - 0.3| = 0.05

Then, we have

\text{TVD} = \frac{1}{2} (0.15 + 0.15 + 0.05 + 0.05) = 0.2

Difficulty: ⭐️⭐️

The average score on this problem was 84%.

Answers: 1, np.arange(1, 50), most_sim_player = player

Let’s try and understand the code provided to us. It appears that we’re looping over the names of all other players, each time computing the TVD between Kelsey Plum’s shot distribution and that player’s shot distribution. If the TVD calculated in an iteration of the for-loop (player_tvd) is less than the previous lowest TVD (lowest_tvd_so_far), the current player (player) is now the most “similar” to Kelsey Plum, and so we store their TVD and name (in most_sim_player).

Before the for-loop, we haven’t looked at any other players, so we don’t have values to store in most_sim_player and lowest_tvd_so_far. On the first iteration of the for-loop, both of these values need to be updated to reflect Kelsey Plum’s similarity with the first player in other_players. This is because, if we’ve only looked at one player, that player is the most similar to Kelsey Plum. most_sim_player is already initialized as an empty string, and we will specify how to “update” most_sim_player in blank (c). For blank (a), we need to pick a value of lowest_tvd_so_far that we can guarantee will be updated on the first iteration of the for-loop. Recall, TVDs range from 0 to 1, with 0 meaning “most similar” and 1 meaning “most different”. This means that no matter what, the TVD between Kelsey Plum’s distribution and the first player’s distribution will be less than 1*, and so if we initialize lowest_tvd_so_far to 1 before the for-loop, we know it will be updated on the first iteration.

• It’s possible that the TVD between Kelsey Plum’s shot distribution and the first other player’s shot distribution is equal to 1, rather than being less than 1. If that were to happen, our code would still generate the correct answer, but lowest_tvd_so_far and most_sim_player wouldn’t be updated on the first iteration. Rather, they’d be updated on the first iteration where player_tvd is strictly less than 1. (We’d expect that the TVDs between all pairs of players are neither exactly 0 nor exactly 1, so this is not a practical issue.) To avoid this issue entirely, we could change if player_tvd < lowest_tvd_so_far to if player_tvd <= lowest_tvd_so_far, which would make sure that even if the first TVD is 1, both lowest_tvd_so_far and most_sim_player are updated on the first iteration.
• Note that we could have initialized lowest_tvd_so_far to a value larger than 1 as well. Suppose we initialized it to 55 (an arbitrary positive integer). On the first iteration of the for-loop, player_tvd will be less than 55, and so lowest_tvd_so_far will be updated.

Then, we need other_players to be an array containing the names of all players other than Kelsey Plum, whose name is stored at position 0 in breakdown.columns. We are told that there are 50 players total, i.e. that there are 50 columns in breakdown. We want to take the elements in breakdown.columns at positions 1, 2, 3, …, 49 (the last element), and the call to np.arange that generates this sequence of positions is np.arange(1, 50). (Remember, np.arange(a, b) does not include the second integer!)

In blank (c), as mentioned in the explanation for blank (a), we need to update the value of most_sim_player. (Note that we only arrive at this line if player_tvd is the lowest pairwise TVD we’ve seen so far.) All this requires is most_sim_player = player, since player contains the name of the player who we are looking at in the current iteration of the for-loop.

Difficulty: ⭐️⭐️⭐️

The average score on this problem was 70%.

Answer: 0.023

Recall, in the solution to the first subpart of this problem, we calculated the absolute differences between the proportions of each category.

• Free Throws: |0.05 - 0.2| = 0.15
• Threes: |0.35 - 0.2| = 0.15
• Midrange: |0.35 - 0.3| = 0.05
• Layups: |0.25 - 0.3| = 0.05

The squared differences between the proportions of each category are computed by squaring the results in the list above (e.g. for Free Throws we’d have (0.05 - 0.2)^2 = 0.15^2). To find the maximum squared difference, then, all we need to do is find the largest of 0.15^2, 0.15^2, 0.05^2, and 0.05^2. Since 0.15 > 0.05, we have that the maximum squared distance is 0.15^2 = 0.0225, which rounds to 0.023.

Difficulty: ⭐️⭐️

The average score on this problem was 85%.

Answers: 10, (shots * possible_points).sum()

To simulate the number of points Kelsey Plum scores in a single game, we need to:

1. Simulate the number of shots she takes of each type (layups, midranges, threes, free throws).
2. Using the simulated distribution in step 1, find the total number of points she scores – specifically, add 2 for every layup, 2 for every midrange, 3 for every three, and 1 for every free throw.

To simulate the number of shots she takes of each type, we use np.random.multinomial. This is because each shot, independently of all other shots, has a 30% chance of being a layup, a 30% chance of being a midrange, and so on. What goes in blank (a) is the number of shots she is taking in total; here, that is 10. shots will be an array of length 4 containing the number of shots of each type - for instance, shots may be np.array([3, 4, 2, 1]), which would mean she took 3 layups, 4 midranges, 2 threes, and 1 free throw.

Now that we have shots, we need to factor in how many points each type of shot is worth. This can be accomplished by multiplying shots with possible_points, which was already defined for us. Using the example where shots is np.array([3, 4, 2, 1]), shots * possible_points evaluates to np.array([6, 8, 6, 1]), which would mean she scored 6 points from layups, 8 points from midranges, and so on. Then, to find the total number of points she scored, we need to compute the sum of this array, either using the np.sum function or .sum() method. As such, the two correct answers for blank (b) are (shots * possible_points).sum() and np.sum(shots * possible_points).

Difficulty: ⭐️⭐️

The average score on this problem was 84%.

Answer: True

The answer is True because of the Central Limit Theorem. Recall, the CLT states that no matter what the population distribution looks like, if you take many repeated samples with replacement, the distribution of the sample means and sample sums will be roughly normal. simulate_points() returns the sum of a sample of size 10 drawn with replacement from a population, and so if we generate many sample sums, the distribution of those sample sums will be roughly normal.

The distribution we are drawing from is the one below.

Difficulty: ⭐️⭐️

The average score on this problem was 78%.

Answer: Options 1, 2, and 4

In the code provided to us, stats is an array containing 10,000 p-values generated by the function fn (note that we are appending stat to stats, and in the line before that we have stat = fn(sim)). In the very last line of the code provided, we have:

win_p_value = np.count_nonzero(stats >= fn([22, 9])) / 10000

If we look closely, we see that we are computing the p-value by computing the proportion of simulated test statistics that were greater than or equal to (>=) the observed statistic. Since a p-value is computed as the proportion of simulated test statistics that were as or more extreme than the observed statistic, here it must mean that “big” test statistics are more extreme.

Remember, the direction that is “extreme” is determined by our alternative hypothesis. Here, the alternative hypothesis is that the Las Vegas Aces win each game with a probability above 60%. As such, the test statistic(s) we choose must be large when the probability that the Aces win a game is high, and small when the probability that the Aces win a game is low. With this all in mind, we can take a look at the 5 options, remembering that arr[0] is the number of simulated wins and arr[1] is the number of simulated losses in a season of 31 games. This means that when the Aces win more than they lose, arr[0] > arr[1], and when they lose more than they win, arr[0] < arr[1].

• Option 1: Here, our test statistic is the ratio of wins to losses, i.e. arr[0] / arr[1]. If the Aces win a lot, the numerator will be larger than the denominator, so this ratio will be large. If the Aces lose a lot, the numerator will be smaller than the denominator, and so this ratio will be small. This is what we want!
• Option 2: Here, our test statistic is the number of wins, i.e. arr[0]. If the Aces win a lot, this number will be large, and if the Aces lose a lot, this number will be small. This is what we want!
• Option 3: Here, our test statistic is the absolute value of the number of wins minus the number of losses, i.e. np.abs(arr[0] - arr[1]). If the Aces win a lot, then arr[0] - arr[1] will be large, and so will np.abs(arr[0] - arr[1]). This seems fine. However, if the Aces lose a lot, then arr[0] - arr[1] will be small (negative), but np.abs(arr[0] - arr[1]) will still be large and positive. This test statistic doesn’t allow us to differentiate when the Aces win a lot or lose a lot, so we can’t use it as a test statistic for our alternative hypothesis.
• Option 4: From the explanation of Option 3, we know that when the Aces win a lot, arr[0] - arr[1] is large. Furthermore, when the Aces lose a lot, arr[0] - arr[1] is small (negative numbers are small in this context). This works!
• Option 5: arr[1] - arr[0] is the opposite of arr[0] - arr[1] in Option 4. When the Aces win a lot, arr[1] - arr[0] is small (negative), and when the Aces lose a lot, arr[1] - arr[0] is large (positive). This is the opposite of what we want, so Option 5 does not work.

Difficulty: ⭐️⭐️

The average score on this problem was 77%.

Answer: Option 4

The distribution visualized in the histogram has the following unique values: -9, -7, -5, -3, …, 17, 19, 21, 23. Crucially, the test statistic whose distribution we’ve visualized can both be positive and negative. Right off the bat, we can eliminate Options 1, 2, and 3:

• Option 1: Invalid. Option 1 is computed by dividing the number of wins (arr[0]) by the number of losses (arr[1]), and that quotient will always be a non-negative number.
• Option 2: Invalid, since the number of wins (arr[0]) will always be a non-negative number.
• Option 3: Invalid, since the absolute value of any real number (np.abs(arr[0] - arr[1]), in this case) will always be a non-negative number.

Now, we must decide between Option 4, whose test statistic is “wins minus losses” (arr[0] - arr[1]), and Option 5, whose test statistic is “losses minus wins” (arr[1] - arr[0]).

First, let’s recap how we’re simulating. In the code provided in the previous subpart, we have the line sim = np.random.multinomial(31, [0.6, 0.4]). Each time we run this line, sim will be set to an array with two elements, the first of which we interpret as the number of simulated wins and the second of which we interpret as the number of simulated losses in a 31 game season. The first number in sim will usually be larger than the second number in sim, since the chance of a win (0.6) is larger than the chance of a loss (0.4). As such, When we compute fn(sim) in the following line, the difference between the wins and losses should typically be positive.

Back to our distribution. Note that the distribution provided in this subpart is centered at a positive number, around 7. Since the difference between wins and losses will typically be positive, it appears that we’ve visualized the distribution of the difference between wins and losses (Option 4). If we instead visualized the difference between losses and wins, the distribution should be centered at a negative number, but that’s not the case.

As such, the correct answer is Option 4.

Difficulty: ⭐️⭐️

The average score on this problem was 86%.

Answer: True

All fn_plus is doing is adding 31 to the output of fn. If we think in terms of pictures, the shape of the distribution of fn_plus looks the same as the distribution of fn, just moved to the right by 31 units. Since the distribution’s shape is no different, the proportion of simulated test statistics that are greater than the observed test statistic is no different either, and so the p-value we calculate with fn_plus is the same as the one we calculate with fn.

Difficulty: ⭐️⭐️⭐️

The average score on this problem was 73%.

Answer: Option 2

The line sim = np.random.multinomial(plum.shape[0], [0.6, 0.4]) assigns sim to an array containing two numbers such that:

• The numbers are randomly chosen each time the line is run
• The numbers always add up to 31

We need to select an option that also creates such an array (or list, in this case). Note that won = plum.get('Won'), a line that is common to all four options, assigns won to a Series with 31 elements, each of which is either True or False (corresponding to the wins and losses that the Las Vegas Aces earned in their season).

Let’s take a look at the line np.count_nonzero(np.random.choice(won, 31, replace=True)), common to the first two options. Here, we are randomly selecting 31 elements from the Series won, with replacement, and counting the number of Trues (since with np.count_nonzero, False is counted as 0). Since we are making our selections with replacement, each selected element has a \frac{22}{31} chance of being True and a \frac{9}{31} chance of being False (since won has 22 Trues and 9 Falses). As such, np.count_nonzero(np.random.choice(won, 31, replace=True)) can be any integer between 0 and 31, inclusive.

Note that if we select without replacement (replace=False) as Option 3 would like us to, then all 31 selected elements would be the same as the 31 elements in won. As a result, np.random.choice(won, 31, replace=False) will always have 22 Trues, just like won, and np.count_nonzero(np.random.choice(won, 31, replace=True)) will always return 22. That’s not random, and so that’s not quite what we’re looking for.

With this all in mind, let’s look at the four options.

• Option 1: Here, each time we call with_rep(), we get a random number between 0 and 31 (inclusive), corresponding to the (random) number of simulated wins. Then, we are assigning sim to be [with_rep(), 31 - with_rep()]. However, it’s not guaranteed that the two calls to with_rep return the same number of wins, so it’s not guaranteed that sum(sim) is 31. Option 1, then, is invalid.
• Option 2: Correct, as we’ll explain below.
• Option 3: As mentioned above, Option 3 uses replace=False, and so without_rep() is always 22 and sim is always [22, 9]. The outcome is not random.
• Option 4: Here, perm() always returns the same number, 22. This is because all we are doing is shuffling the entries in the won Series, but we aren’t changing the number of wins (Trues) and losses (Falses). As a result, w is always 22 and sim is always [22, 9], making this non-random, just like in Option 3.

By the process of elimination, Option 2 must be the correct choice. It is similar to Option 1, but it only calls with_rep once and “saves” the result to the name w. As a result, w is random, and w and 31 - w are guaranteed to sum to 31.

⚠️ Note: It turns out that none of these options run a valid hypothesis test, since the null hypothesis was that the Las Vegas Aces win 60% of their games but none of these simulation strategies use 60% anywhere (instead, they use the observation that the Aces actually won 22 games). However, this subpart was about the sampling strategies themselves, so this mistake from our end doesn’t invalidate the problem.

Difficulty: ⭐️⭐️⭐️

The average score on this problem was 70%.

Answers: Options 2, 3, and 4

• Option 1: As explained in the solution to the previous subpart, if the two calls to with_rep evaluate to different numbers (entirely possible, since it is random), then sum(sim) will not be 31.
• Option 2: Here, sim is defined in terms of some w. Specifically, w is some number between 0 and 31 and sim is [w, 31 - w], so sum(sim) is the same as w + 31 - w, which is always 31.
• Option 3: In Option 3, sim is always [22, 9], and sum(sim) is always 31.
• Option 4: Same as Option 3.

Difficulty: ⭐️⭐️⭐️

The average score on this problem was 72%.

Answers: 'Won', 'AST'

To compute the number of assists Kelsey Plum averages in winning and losing games, we need to group by 'Won'. Once doing so, and using the .mean() aggregation method, we need to access elements in the 'AST' column.

The second argument to diff_in_group_means, group_col, is the column we’re grouping by, and so blank (a) must be filled by 'Won'. Then, the second argument, num_col, must be 'AST'.

Note that after extracting the Series containing the average number of assists in wins and losses, we are returning the value with the index False (“loss”) minus the value with the index True (“win”). So, throughout this problem, keep in mind that we are computing “losses minus wins”. Since our observation was that she averaged 0.61 more assists in wins than in losses, it makes sense that diff_in_group_means(plum, 'Won', 'AST') is -0.61 (rather than +0.61).

Difficulty: ⭐️

The average score on this problem was 94%.

Answer: diff_in_group_means(shuffled, 'Won', 'AST')

As we saw in the previous subpart, diff_in_group_means(shuffled, 'Won', 'AST') computes the observed test statistic, which is -0.61. There is no randomness involved in the observed test statistic; each time we run the line diff_in_group_means(shuffled, 'Won', 'AST') we will see the same result, so this cannot be used for simulation.

To perform a permutation test here, we need to simulate under the null by randomly assigning assist counts to groups; here, the groups are “win” and “loss”.

• Option 2: Here, assist counts are shuffled and the group names are kept in the same order. The end result is a random pairing of assists to groups.
• Option 3: Here, the group names are shuffled and the assist counts are kept in the same order. The end result is a random pairing of assist counts to groups.
• Option 4: Here, both the group names and assist counts are shuffled, but the end result is still the same as in the previous two options.

As such, Options 2 through 4 are all valid, and Option 1 is the only invalid one.

Difficulty: ⭐️⭐️⭐️

The average score on this problem was 68%.

Answer: Under the assumption that Kelsey Plum’s number of assists in winning games and in losing games come from the same distribution, and that she wins 22 of the 31 games she plays, the chance of her averaging at least 0.61 more assists in wins than losses is roughly a quarter. (Option 3)

First, we should note that the area to the left of the red line (a quarter) is the p-value of our hypothesis test. Generally, the p-value is the probability of observing an outcome as or more extreme than the observed, under the assumption that the null hypothesis is true. The direction to look in depends on the alternate hypothesis; here, since our alternative hypothesis is that the number of assists Kelsey Plum makes in winning games is higher on average than in losing games, a “more extreme” outcome is where the assists in winning games are higher than in losing games, i.e. where \text{(assists in wins)} - \text{(assists in losses)} is positive or where \text{(assists in losses)} - \text{(assists in wins)} is negative. As mentioned in the solution to the first subpart, our test statistic is \text{(assists in losses)} - \text{(assists in wins)}, so a more extreme outcome is one where this is negative, i.e. to the left of the observed statistic.

Let’s first rule out the first two options.

• Option 1: This option states that the probability that the null hypothesis (the number of assists she makes in winning and losing games comes from the same distribution) is true is roughly a quarter. However, the p-value is not the probability that the null hypothesis is true.
• Option 2: The significance level is the formal name for the p-value “cutoff” that we specify in our hypothesis test. There is no cutoff mentioned in the problem. The observed significance level is another name for the p-value, but Option 2 did not contain the word observed.

Now, the only difference between Options 3 and 4 is the inclusion of “at least” in Option 3. Remember, to compute a p-value we must compute the probability of observing something as or more extreme than the observed, under the null. The “or more” corresponds to “at least” in Option 3. As such, Option 3 is the correct choice.

Difficulty: ⭐️⭐️⭐️

The average score on this problem was 70%.

Answer: False

The area of a density histogram is 1. The area of the histogram drawn in the previous subpart is much larger than 1. In fact, the area of this histogram is in the hundreds or thousands; you can draw a rectangle stretching from -1 to 1 on the x-axis and 0 to 300 on the y-axis that has area 2 \cdot 300 = 600, and this rectangle is much smaller than the larger histogram.

Difficulty: ⭐️⭐️

The average score on this problem was 82%.

Answer: \frac{1}{2^{31}}

First, let’s understand what df_choice does. It takes in one input, df. The line np.random.choice([True, False], df.shape[0], replace=True) evaluates to an array such that:

• There is one element for every row in the input DataFrame or array df (so if df has 31 rows, the output array will have length 31)
• Each element is equally likely to be True or False, since the sequence we are selecting from is [True, False] and we are selecting with replacement

So np.random.choice([True, False], df.shape[0], replace=True) is an array the same length as df, with each element randomly set to True or False. Note that there is a \frac{1}{2} chance the first element is True, a \frac{1}{2} chance the second element is True, and so on.

Then, df[np.random.choice([True, False], df.shape[0], replace=True)] is using Boolean indexing to keep only the rows in df where the array np.random.choice([True, False], df.shape[0], replace=True) contains the value True. So, the function df_choice returns a DataFrame containing somewhere between 0 and df.shape[0] rows. Note that there is a \frac{1}{2} chance that the new DataFrame contains the first row from df, a \frac{1}{2} chance that the new DataFrame contains the second row from df, and so on.

In this question, the only input ever given to df_choice is plum, which has 31 rows.

In this subpart, we’re asked for the probability that df_choice(plum) is an empty DataFrame. There are 31 rows, and each of them have a \frac{1}{2} chance of being included in the output, and so a \frac{1}{2} chance of being missing. So, the chance that they are all missing is:

\begin{aligned} P(\text{empty DataFrame}) &= P(\text{all rows missing}) \\ &= P(\text{row 0 missing and row 1 missing and ... and row 30 missing}) \\ &= P(\text{row 0 missing}) \cdot P(\text{row 1 missing}) \cdot ... \cdot P(\text{row 30 missing}) \\ &= \frac{1}{2} \cdot \frac{1}{2} \cdot ... \cdot \frac{1}{2} \\ &= \boxed{\frac{1}{2^{31}}} \end{aligned}

Difficulty: ⭐️⭐️

The average score on this problem was 83%.

Answer: \frac{31}{2^{31}}

In order for the resulting DataFrame to have 30 rows, exactly 1 row must be missing, and the other 30 must be present.

To start, let’s consider one row in particular, say, row 7. The probability that row 7 is missing is \frac{1}{2}, and the probability that rows 0 through 6 and 8 through 30 are all present is \frac{1}{2} \cdot \frac{1}{2} \cdot ... \cdot \frac{1}{2} = \frac{1}{2^{30}} using the logic from the previous subpart. So, the probability that row 7 is missing AND all other rows are present is \frac{1}{2} \cdot \frac{1}{2^{30}} = \frac{1}{2^{31}}.

Then, in order for there to be 30 rows, either row 0 must be missing, or row 1 must be missing, and so on:

\begin{aligned} P(\text{exactly one row missing}) &= P(\text{only row 0 is missing or only row 1 is missing or ... or only row 30 is missing}) \\ &= P(\text{only row 0 is missing}) + P(\text{only row 1 is missing}) + ... + P(\text{only row 30 is missing}) \\ &= \frac{1}{2^{31}} + \frac{1}{2^{31}} + ... + \frac{1}{2^{31}} \\ &= \boxed{\frac{31}{2^{31}}} \end{aligned}

Difficulty: ⭐️⭐️⭐️⭐️

The average score on this problem was 48%.

Answer: True

An important realization to make is that all subsets of the rows in plum are equally likely to be returned by df_choice(plum), and they all have probability \frac{1}{2^{31}}. For instance, one subset of plum is the subset where rows 2, 5, 8, and 30 are missing, and the rest are all present. The probability that this subset is returned by df_choice(plum) is \frac{1}{2^{31}}.

This is true because for each individual row, the probability that it is present or missing is the same – \frac{1}{2} – so the probability of any subset is a product of 31 \frac{1}{2}s, which is \frac{1}{2^{31}}. (The answer to the previous subpart was not \frac{1}{2^{31}} because it was asking about multiple subsets – the subset where only row 0 was missing, and the subset where only row 1 was missing, and so on).

So, the probability that df_choice(plum) consists of just row 0 is \frac{1}{2^{31}}, and this is the same as the answer to the first subpart (\frac{1}{2^{31}}); in both situations, we are calculating the probability of one specific subset.

Difficulty: ⭐️⭐️⭐️

The average score on this problem was 63%.

Answer: \frac{1}{2^{25}}

Here, we’re not being asked for the probability of one specific subset (like the subset containing just row 0); rather, we’re being asked for the probability of various different subsets, so our calculation will be a bit more involved.

We can break our problem down into four pieces. We can find the probability that there are 0 rows, 1 row, 30 rows, and 31 rows individually, and add these probabilities up, since only one of them can happen at a time (it’s impossible for a DataFrame to have both 1 and 30 rows at the same time; these events are “mutually exclusive”). It turns out we’ve already calculated two of these probabilities:

• From the first subpart, the probability that 0 rows are returned is \frac{1}{2^{31}}. This corresponds to a single subset, the subset where all rows are missing.
• From the second subpart, the probability that 30 rows are returned is \frac{31}{2^{31}}.

The other two possibilities are symmetric with the above two!

• The probability that 31 rows are returned is the same as the probability that 0 rows are returned, since the probability of a row being missing and a row being present is the same. This is \frac{1}{2^{31}}.
• The probability that 1 row is returned is the same as the probability that 1 row is missing, i.e. the probability that 30 rows are returned. This, from the second subpart, is \frac{31}{2^{31}}.

Putting it all together, we have:

\begin{aligned} P(\text{number of returned rows is 0, 1, 30, or 31}) &= P(\text{0 rows are returned}) + P(\text{1 row is returned}) + P(\text{30 rows are returned}) + P(\text{31 rows are returned}) \\ &= \frac{1}{2^{31}} + \frac{31}{2^{31}} + \frac{31}{2^{31}} + \frac{1}{2^{31}} \\ &= \frac{1 + 31 + 31 + 1}{2^{31}} \\ &= \frac{64}{2^{31}} \\ &= \frac{2^6}{2^{31}} \\ &= \frac{1}{2^{31 - 6}} \\ &= \boxed{\frac{1}{2^{25}}} \end{aligned}

Difficulty: ⭐️⭐️⭐️⭐️

The average score on this problem was 35%.

Answer: 'Player'

Ideally, the index of a DataFrame is unique, so that we can use it to “identify” the rows. Here, each row is about a player, so 'Player' should be the index. 'Player' is the only column that is likely to be unique; it is possible that two players have the same name, but it’s still a better choice of index than the other three options, which are definitely not unique.

Difficulty: ⭐️

The average score on this problem was 95%.

Answer: 1

The median of a distribution is the value that is “halfway” through the distribution, i.e. the value such that half of the values in the distribution are larger than it and half the values in the distribution are smaller than it.

Visually, we’re looking for the location on the x-axis where we can draw a vertical line that splits the area of the histogram in half. While it’s impossible to tell the exact median of the distribution, since we don’t know how the values are distributed within the bars, we can get pretty close by using this principle.

Immediately, we can rule out 0.5, 0.75, 1.5, and 1.75, since they are too far from the “center” of the distribution (imagine drawing vertical lines at any of those points on the x-axis; they don’t split the distribution’s area in half). To decide between 1 and 1.25, we can use the fact that the distribution is right-skewed, meaning that its mean is larger than its median (intuitively, the mean is dragged in the direction of the tail, which is to the right). This means that the median should be less than the mean. We are given that the mean of the distribution is 1.25, so the median should be 1.

Difficulty: ⭐️⭐️⭐️

The average score on this problem was 73%.

Answer: 0.5

To convert a value x to standard units (denoted by x_{\text{su}}), we use the following formula:

x_{\text{su}} = \frac{x - \text{mean of }x}{\text{SD of }x}

Let’s look at the first line given to us: In “original units”, Sabrina Ionescu had 3.5 turnovers per game. In standard units, her turnovers per game is 3.

Substituting the information we know into the above equation gives us:

3 = \frac{3.5 - 1.25}{\text{SD of }x}

In order to convert future values from original units to standard units, we’ll need to know \text{SD of }x, which we don’t currently but can obtain by rearranging the above equation. Doing so yields

\text{SD of }x = \frac{3.5-1.25}{3} = \frac{2.25}{3} = 0.75

Now, let’s look at the second line we’re given: In standard units, Sami Whitcomb’s turnovers per game is -1. How many turnovers per game did Sami Whitcomb have in original units? Round your answer to 3 decimal places.

We have all the information we need to convert Sami Whitcomb’s turnovers per game from standard units to original units! Plugging in the values we know gives us:

\begin{aligned} x_{\text{su}} &= \frac{x - \text{mean of }x}{\text{SD of }x} \\ -1 &= \frac{x - 1.25}{0.75} \\ -0.75 &= x - 1.25 \\ 1.25 - 0.75 &= x \\ x &= \boxed{0.5} \end{aligned}

Thus, in original units, Sami Whitcomb averaged 0.5 turnovers per game.

Difficulty: ⭐️⭐️

The average score on this problem was 87%.

Answer: -1.667

The smallest possible number of turnovers per game in original units is 0 (which a player would have if they never had a turnover – that would mean they’re really good!). To find the smallest possible turnovers per game in standard units, all we need to do is convert 0 from original units to standard units. This will involve our work from the previous subpart.

\begin{aligned} x_{\text{su}} &= \frac{x - \text{mean of }x}{\text{SD of }x} \\ &= \frac{0 - 1.25}{0.75} \\ &= -\frac{1.25}{0.75} \\ &= -\frac{5}{3} = \boxed{-1.667} \end{aligned}

Difficulty: ⭐️⭐️

The average score on this problem was 82%.

Answer: 5.4

We need to find and use the regression line to find the predicted y for an x of 27. There are two ways to proceed:

1. Use the regression line in standard units. To do this, we’d need to convert 27 from original units to standard units, use the regression line y_\text{su} = r \cdot x_\text{su}, and convert the output back to original units.
2. Use the regression line in original units. To do this, we’d need to find the slope m and intercept b in the regression line y = mx + b, using the formulas m = r \cdot \frac{\text{SD of }y }{\text{SD of }x} and b = \text{mean of }y - m \cdot \text{mean of }x.

Both solutions work; for the sake of completeness, we’ll show both. Recall, r is the correlation coefficient between x and y, which we are told is 0.65.

Solution 1:

First, we need to convert 27 points per game to standard units. Doing so yields

x_{\text{su}} = \frac{x - \text{mean of }x}{\text{SD of }x} = \frac{27 - 7}{5} = 4

Per the regression line, y_\text{su} = r \cdot x_\text{su}, we have y_\text{su} = 0.65 \cdot 4 = 2.6, which is Tina Charles’ predicted assists per game in standard units. All that’s left is to convert this value back to original units:

\begin{aligned} y_{\text{su}} &= \frac{y - \text{mean of }y}{\text{SD of }y} \\ 2.6 &= \frac{y - 1.5}{1.5} \\ 2.6 \cdot 1.5 + 1.5 &= y \\ y &= \boxed{5.4} \end{aligned}

So, the regression line predicts Tina Charles will have 5.4 assists per game (in original units).

Solution 2:

First, we need to find the slope m and intercept b:

m = r \cdot \frac{\text{SD of }y }{\text{SD of }x} = 0.65 \cdot \frac{1.5}{5} = 0.195

b = \text{mean of }y - m \cdot \text{mean of }x = 1.5 - 0.195 \cdot 7 = 0.135

Then,

y = mx + b \implies y = 0.195 \cdot 27 + 0.135 = \boxed{5.4}

So, once again, the regression line predicts Tina Charles will have 5.4 assists per game.

Note: The numbers in this problem may seem ugly, but students taking this exam had access to calculators since this exam was online. It also turns out that the numbers were easier to work with in Solution 1 over Solution 2; this was intentional.

Difficulty: ⭐️⭐️

The average score on this problem was 81%.

Answer: -3.3

Residuals are defined as follows:

\text{residual} = \text{actual } y - \text{predicted }y

2.1 - 5.4 = -3.3, which gives us our answer.

Note: Many students answered 3.3. Pay attention to the order of the calculation!

Difficulty: ⭐️⭐️

The average score on this problem was 82%.

Answers:

• The point (0, 0) is guaranteed to be on the regression line when both x and y are in standard units (Option 1).
• The point (7, 1.5) is guaranteed to be on the regression line when both x and y are in original units (Option 4).

The main idea being assessed here is the fact that the point (\text{mean of }x, \text{mean of }y) always lies on the regression line. Indeed, in original units, 7 is the average x (PPG) and 1.5 is the average y (APG); this information was provided to us at the start of the problem. The nuance behind this problem lies in the units that are being used in the regression line.

When the regression line is in standard units:

• In standard units, 0 means “0 standard deviations above the average”, i.e. 0 means “average”. When the regression line is in standard units, we have y_\text{su} = r \cdot x_\text{su}. If x is average, i.e. if x_\text{su} = 0, then y_\text{su} = r \cdot x_\text{su} = r \cdot 0 = 0, regardless of what r is. So the point (0, 0) is on the regression line when both x and y are in standard units, meaning that Option 1 is correct.
• The point (7, 1.5) is not on the regression line when x and y are in standard units. We know this because in this problem r = 0.65, and if x_\text{su} = 7, then y_\text{su} = r \cdot x_\text{su} = 0.65 \cdot 7 = 4.55 \neq 1.5. This means that Option 3 is incorrect.

When the regression line is in original units:

• In original units, the average x is 7 and the average y is 1.5. From class, we may remember that this automatically means that (7, 1.5) is on the regression line in original units. If we didn’t remember that, we can look to the formula for the slope m and intercept b in y = mx + b. The formula for the slope is actually not relevant here; what’s relevant is the fact that b = \text{mean of }y - m \cdot \text{mean of }x. Substituting the formula for b into y = mx + b yields

y = mx + b = mx + \text{mean of }y - m \cdot \text{mean of }x

• If x = \text{mean of }x, then the above simplifies to: y = m \cdot \text{mean of }x + \text{mean of }y - m \cdot \text{mean of }x = \text{mean of }y, meaning that (\text{mean of }x, \text{mean of }y) — which is (7, 1.5) in this case — is on the regression line in original units, so Option 4 is correct.
• In the above equation, if x = 0, then y = \text{mean of }y - m \cdot \text{mean of }x, which in this case simplifies to 1.5 - 0.195 \cdot 7 = 0.135 \neq 0. This means that (0, 0) is not on the regression line in original units and Option 2 is incorrect.

Difficulty: ⭐️⭐️

The average score on this problem was 87%.

Answer: a < c

The formula for the slope of the regression line is m = r \cdot \frac{\text{SD of }y}{\text{SD of }x}. Note that the correlation coefficient r is symmetric, meaning that the correlation between x and y is the same as the correlation between y and x.

In the two regression lines mentioned in this problem, we have

\begin{aligned} a &= r \cdot \frac{\text{SD of assists per game}}{\text{SD of points per game}} \\ c &= r \cdot \frac{\text{SD of points per game}}{\text{SD of assists per game}} \end{aligned}

We’re told in the problem that the SD of points per game is 5 and the SD of assists per game is 1.5. So, a = r \cdot \frac{1.5}{5} and c = r \cdot \frac{5}{1.5}; since \frac{1.5}{5} < \frac{5}{1.5}, a < c.

Difficulty: ⭐️⭐️⭐️

The average score on this problem was 74%.

Answer: 92.284\%

Recall, Chebyshev’s inequality states that the proportion of values within z standard deviations of the mean is at least 1 - \frac{1}{z^2}.

To approach the problem, we’ll start by converting 25 points per game to standard units. Doing so yields \frac{25 - 7}{5} = 3.6. This means that 25 is 3.6 standard deviations above the mean. The value 3.6 standard deviations below the mean is 7 - 3.6 \cdot 5 = -11, so when we use Chebyshev’s inequality with z = 3.6, we will get a lower bound on the proportion of values between -11 and 25. However, as the question tells us, points per game must be non-negative, so in this case the proportion of values between -11 and 25 is the same as the proportion of values between 0 and 25 (i.e. the proportion of values less than or equal to 25).

When z = 3.6, we have 1 - \frac{1}{z^2} = 1 - \frac{1}{3.6^2} = 0.922839, which as a percentage rounded to three decimal places is 92.284\%. Thus, at least 92.284\% scored 25 or fewer points per game.

Difficulty: ⭐️⭐️⭐️⭐️

The average score on this problem was 46%.

Answer: 85

For reference, recall that we find the pth percentile of a collection of n numbers as follows:

1. Sort the collection in increasing order.
2. Define h to be p\% of n:

h = \frac{p}{100} \cdot n

3. If h is an integer, define k = h. Otherwise, let k be the smallest integer greater than h.

4. Take the kth element of the sorted collection (start counting from 1, not 0).

To start, it’s worth emphasizing that there are n = 14 numbers in aces total. 14.24 is at position 12 (when the positions are numbered 1 through 14).

Let’s try and find a value of p such that 14.24 is the pth percentile. To do so, we might try and find what “percentage” of the way through the distribution 14.24 is; doing so gives \frac{12}{14} = 85.71\%. If we follow the process outlined above with p = 85, we get that h = \frac{85}{100} \cdot 14 = 11.9 and thus k = 12, meaning that the 85th percentile is the number at position 12, which 14.24.

Let’s see what happens when we try the same process with p = 86. This time, we have h = \frac{86}{100} \cdot 14 = 12.04 and thus k = 13, meaning that the 86th percentile is the number at position 13, which is 14.81.

This means that the value of q is 85 – the 85th percentile is 14.24, while the 86th percentile is 14.81.

Difficulty: ⭐️⭐️⭐️

The average score on this problem was 57%.

Answers:

• The sample size is season.shape[0], when it should be small_season.shape[0]
• Sampling is currently being done without replacement, when it should be done with replacement

Here, our goal is to bootstrap from small_season. When bootstrapping, we sample with replacement from our original sample, with a sample size that’s equal to the original sample’s size. Here, our original sample is small_season, so we should be taking samples of size small_season.shape[0] from it.

Option 1 is incorrect; season has nothing to do with this problem, as we are bootstrapping from small_season.

Difficulty: ⭐️

The average score on this problem was 95%.

Answer: Currently it is taking the mean of the 'PPG' column in small_season, when it should be taking the mean of the 'PPG' column in resample

The current implementation of Line 2 doesn’t use the resample at all, when it should. If we were to leave Line 2 as it is, all of the values in boot_means would be identical (and equal to the mean of the 'PPG' column in small_season).

Option 1 is incorrect since our bootstrapping procedure is independent of season. Option 3 is incorrect because .mean() is a valid Series method.

Difficulty: ⭐️

The average score on this problem was 98%.

Answers:

• The result of calling np.append is not being reassigned to boot_means, so boot_means will be an empty array after running this procedure
• new_mean is not a defined variable name, and should be replaced with resample_mean

np.append returns a new array and does not modify the array it is called on (boot_means, in this case), so Option 1 is a necessary fix. Furthermore, Option 3 is a necessary fix since new_mean wasn’t defined anywhere.

Option 2 is incorrect; if np.append were outside of the for-loop, none of the 10,000 resampled means would be saved in boot_means.

Difficulty: ⭐️

The average score on this problem was 94%.

Answer: Option 3

The key to this problem is knowing to use the Central Limit Theorem. Specifically, we know that if we collect many samples from a population with replacement, then the distribution of the sample means will be roughly normal with:

• a mean that is equal to the mean of the population
• a standard deviation that is \frac{\text{SD of population}}{\sqrt{\text{sample size}}}

Here, the “population” is small_season, because that is the sample we’re repeatedly (re)sampling from. While season is actually the population, it is not seen at all in the bootstrapping process, so it doesn’t directly influence the distribution of the bootstrapped sample means.

The mean of small_season is 9, and so is the distribution of bootstrapped sample means. The standard deviation of small_season is 4, so the square root law, the standard deviation of the distribution of bootstrapped sample means is \frac{4}{\sqrt{36}} = \frac{4}{6} = \frac{2}{3}.

The answer now boils down to choosing the histogram that looks roughly normally distributed with a mean of 9 and a standard deviation of \frac{2}{3}. Options 1 and 4 can be ruled out right away since their means seem to be smaller than 9. To decide between Options 2 and 3, we can use the inflection point rule, which states that in a normal distribution, the inflection points occur at one standard deviation above and one standard deviation below the mean. (An inflection point is when a curve changes from opening upwards to opening downwards.) See the picture below for more details.

Option 3 is the only distribution that appears to be centered at 9 with a standard deviation of \frac{2}{3} (0.7 is close to \frac{2}{3}), so it must be the empirical distribution of the bootstrapped sample means.

Difficulty: ⭐️⭐️⭐️⭐️

The average score on this problem was 42%.

Answer: (left_b + right_b) / 2 is not necessarily equal to the mean points per game in small_season, but is close.

Normal-based confidence intervals are of the form [\text{mean} - \text{something}, \text{mean} + \text{something}]. In such confidence intervals, it is the case that the average of the left and right endpoints is exactly the mean of the distribution used to compute the interval.

However, the confidence interval we’ve created is not normal-based, rather it is bootstrap-based! As such, we can’t say that anything is exactly true; this rules out Options 1 and 3.

Our 95% confidence interval was created by taking the middle 95% of bootstrapped sample means. The distribution of bootstrapped sample means is roughly normal, and the normal distribution is symmetric (the mean and median are both equal, and represent the “center” of the distribution). This means that the middle of our 95% confidence interval should be roughly equal to the mean of the distribution of bootstrapped sample means. This implies that Option 4 is correct; the difference between Options 2 and 4 is that Option 4 uses small_season, which is the sample we bootstrapped from, while Option 2 uses season, which was not accessed at all in our bootstrapping procedure.

Difficulty: ⭐️⭐️⭐️

The average score on this problem was 62%.

Answer: [7.667, 10.333]

In a normal distribution, roughly 95% of values are within 2 standard deviations of the mean. The CLT tells us that the distribution of sample means is roughly normal, and in subpart 4 of this problem we already computed the SD of the distribution of sample means to be \frac{2}{3}.

So, our normal-based 95% confidence interval is computed as follows:

\begin{aligned} &[\text{mean of sample} - 2 \cdot \text{SD of distribution of sample means}, \text{mean of sample} + 2 \cdot \text{SD of distribution of sample means}] \\ &= [9 - 2 \cdot \frac{4}{\sqrt{36}}, 9 + 2 \cdot \frac{4}{\sqrt{36}}] \\ &= [9 - \frac{4}{3}, 9 + \frac{4}{3}] \\ &\approx \boxed{[7.667, 10.333]} \end{aligned}

Difficulty: ⭐️⭐️

The average score on this problem was 87%.

Answer: The 95% confidence interval we created in the previous subpart did not contain the true mean points per game, but if we collected many original samples and constructed many 95% confidence intervals, then roughly 95% of them would contain the true mean points per game.

In a confidence interval, the confidence level gives us a level of confidence in the process used to create the confidence interval. If we repeat the process of collecting a sample from the population and using the sample to construct a c% confidence interval for the population mean, then roughly c% of the intervals we create should contain the population mean. Option 4 is the only option that corresponds to this interpretation; the others are all incorrect in different ways.

Difficulty: ⭐️⭐️

The average score on this problem was 87%.

Answers:

• Drinking Gatorade causes a player to gain Instagram followers.
• Deleting Twitter causes a player to gain Instagram followers.

The key to this problem is understanding the nature of randomized controlled trials (RCTs). To run an RCT, we must be able to randomly assign our test subjects to either a treatment or control group, and apply some treatment only to the treatment group. With that in mind, let’s look at the four options.

• Option 1: Here, the treatment would be winning two games in a row. However, this is not something we could apply to a treatment group, since we can’t control whether or not basketball players win games (let alone two games in a row). As such, this is not an RCT we could run.
• Option 2: Here, the treatment would be drinking Gatorade. We indeed could have a treatment group drink Gatorade; this is no different than having a treatment group take some form of medicine. This is an RCT we could run.
• Option 3: Here, the treatment would be playing for the Las Vegas Aces. Similar to Option 1, this is not an RCT we could run, since we can’t control which teams players play for.
• Option 4: Here, the treatment would be deleting Twitter. While they may not like it, we could have players participating in our study delete their Twitter accounts, so this is an RCT we could run.

Note, our assessment above did not look at the outcome, gaining Instagram followers, at all. While to us it may seem unlikely that drinking Gatorade causes a player to gain Instagram followers, that doesn’t mean we can’t run an RCT to check.

Difficulty: ⭐️⭐️

The average score on this problem was 77%.