Correction of 1st exam 2024/25

NES—Programmiertechniken (Introduction to scientific programming)

31.01.2025

Please work through the following tasks. You have 75 minutes to complete the exam.
Make sure you executed all relevant code cells and save the notebook before the end of the exam.

This is a correction of the exam questions. Like always in programming, there are many ways to achieve similar results. My suggestions here follow our coding exercises discussed in class.

If you want to solve this exam as an exercise, download the uncorrected notebook.

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt

Question 1: Theory (5 points)

(a) Multiple-choice

To tick correct answers, replace - [ ] with - [x]

Careful: Wrongly ticked answers will give negative points!

Which of the following statements are true? (2.5 points)

(b) Rapid-response

  • Example: What programming language did our course focus on?
    • Answer: Python

Careful: Wrong answers will give negative points. No answers will yield 0 points (i.e., no negative and no positive points)

Answer the following prompts with one word only! (2.5 points)

  1. Name one package and environment manager for Python!
    • Answer: conda
  2. What file ending do common Python scripts have?
    • Answer: .py
  3. Which keyword is used to skip the current iteration in a loop?
    • Answer: continue
  4. Name one immutable data type in Python!
    • Answer: tuple, string
  5. What is the data structure called that provides a name for each row in a Pandas DataFrame?
    • Answer: Index

Question 2: A wind turbine on top of Piz Buin

Imagine you had to place a wind turbine on top of Piz Buin, Vorarlberg’s highest mountain. A turbine that always faces the same direction and can’t turn into the wind. In which direction should it look? You will work with wind data from Piz Buin for all of 2024 to make a decision.

Part I: Initial data exploration (6 points)

  1. Read the dataset wind_PizBuin.csv into a DataFrame called df and create a DatetimeIndex with the column datetime. (2 points)
df = pd.read_csv("wind_PizBuin.csv", parse_dates=True, index_col='datetime')
df
wind_speed wind_direction
datetime
2024-01-01 00:00:00 4.1 15
2024-01-01 01:00:00 4.4 351
2024-01-01 02:00:00 5.4 360
2024-01-01 03:00:00 3.2 360
2024-01-01 04:00:00 3.8 17
... ... ...
2024-12-31 19:00:00 1.5 300
2024-12-31 20:00:00 1.4 293
2024-12-31 21:00:00 1.5 263
2024-12-31 22:00:00 1.8 233
2024-12-31 23:00:00 2.0 225

8784 rows × 2 columns

  1. Create a histogram of the column 'wind_direction'. (0.5 points)
df['wind_direction'].hist()

  1. Add a column to df named wind_sector, which should be 'S' (for South) when 270 > wind_direction >= 90, and 'N' (for North) otherwise. Do not use a loop to solve this task. (2 points)
df['wind_sector'] = 'N'
df.loc[(df['wind_direction'] >= 90) & (df['wind_direction'] < 270), 'wind_sector'] = 'S'
df
wind_speed wind_direction wind_sector
datetime
2024-01-01 00:00:00 4.1 15 N
2024-01-01 01:00:00 4.4 351 N
2024-01-01 02:00:00 5.4 360 N
2024-01-01 03:00:00 3.2 360 N
2024-01-01 04:00:00 3.8 17 N
... ... ... ...
2024-12-31 19:00:00 1.5 300 N
2024-12-31 20:00:00 1.4 293 N
2024-12-31 21:00:00 1.5 263 S
2024-12-31 22:00:00 1.8 233 S
2024-12-31 23:00:00 2.0 225 S

8784 rows × 3 columns

  1. Using the .groupby() method, compute the median wind_speed and median wind_direction for each of the two wind_sectors. (1.5 points)
df.groupby('wind_sector').median()
wind_speed wind_direction
wind_sector
N 4.8 305.0
S 5.4 166.0

From the histogram we see that the wind generally blows from two directions. It tends to be slightly stronger for the southerly wind, but we still can’t make a decision. What if the turbine is best placed in between these two directions? We need more detailed analyses..

Part II: Grid search (9 points)

Using a grid search, identify in which direction the wind turbine should ideally look to produce most energy. To do so, perform the following tasks:

  1. Import the wind_module. (0.5 points)
import wind_module
  1. Define a function named power2energy that computes energy from an array of wind power and the time sampling dt. (2.5 points)
    • Set the default value for dt to 1 (which aligns with our hourly dataset).
    • Reminder: \(E = \sum P \cdot \Delta t\), where \(E\) is energy, \(P\) is power, and \(\Delta t\) is dt.
def power2energy(power, dt=1):
    energy = np.sum(power) * dt
    return energy
  1. Define a numpy array turbine_directions with angles from 0 to 355 in steps of 5 degrees,
    and initialize an array energies of the same shape with np.nan. (1 point)
turbine_directions = np.arange(0, 360, 5)
energies = np.full(turbine_directions.shape, np.nan)
  1. Iterate over the array turbine_directions and
    • compute the time series of power produced by the wind turbine using the function compute_wind_power() in the wind_module. You will have to provide wind speed and direction from the DataFrame df (Part I), and the turbine direction.
    • convert the time series of power to energy and store it at the relevant location within energies.
    (3 points)
for i, direction in enumerate(turbine_directions):
    power = wind_module.compute_wind_power(df['wind_speed'], df['wind_direction'], direction)
    energy = power2energy(power)
    energies[i] = energy
  1. Normalize the array energies with its maximum value and store the result in a new array named energies_normalized. (1 point)
energies_normalized = energies/np.max(energies)
  1. Which turbine direction produced most energy? (1 point)
    Write a short code snippet to find the answer. Here is a visualization to verify whether your result makes sense:

turbine_directions[np.argmax(energies_normalized)]
np.int64(165)

Part III: Working with and plotting timeseries (12 points)

  1. How many unique day_of_week does the DatetimeIndex of the DataFrame df contain? Write a code snippet to find the answer! (1.5 points)
unique_days_of_week = df.index.day_of_week.unique()
unique_days_of_week.shape[0]
7
  1. Subset the DataFrame df to the period of 29th of January 2024 – 4th of February 2024 and name the resulting DataFrame dfs.
    Make sure you do this in a way that avoids any potential side effects (or SettingWithCopyWarnings later in the code)! (1.5 points)
dfs = df["2024-01-29":"2024-02-04"].copy()
  1. Using dfs, create a plot as closely as possible to the following one:
    • one curve represents the wind speed as provided in the DataFrame
    • one curve shows the 5 hour resampled mean wind speed (Tip: Use the frequency '5h' to achieve this!)
    • one curve displays a moving average with a 5 hour window. Implement the moving average with a for-loop that fills a new column named dfs['wind_ma_5h']!
    (9 points)

dfs['wind_ma_5h'] = np.nan
for i in range(2, dfs.shape[0]-2):
    dfs.loc[dfs.index[i], 'wind_ma_5h'] = dfs.loc[dfs.index[i-2]:dfs.index[i+2], 'wind_speed'].mean()
    
dfs_5h_mean = dfs['wind_speed'].resample('5h').mean()
f, ax = plt.subplots()
dfs['wind_speed'].plot(ax=ax, label="hourly")
dfs_5h_mean.plot(ax=ax, label="5h resampled mean")
dfs['wind_ma_5h'].plot(ax=ax, color='black', label="5h moving average")
ax.legend()
ax.set_xlabel("")
ax.set_ylabel("Wind speed (m/s)")
ax.grid(linestyle=":")
plt.show()

Question 3: Debug a code snippet!

The following code cell contains 1 syntax error and 3 semantic errors.
Fix them! (4.5 points)

# Code block with errors
keywords = ["Python", "programming", "exam"]
letter   = "o"

count_total = 0
count_per_keyword = 0

for word in keywords:
    for char in word:
        if char is letter:
            count_per_keyword += 1
    print(f"{word:14}: "{char}" appears {count_per_keyword} times.")
    count_total += count_per_keyword
    count_per_keyword = 0

print("")
print("In total, letter {letter} appears {count_total} times.")
  Cell In[18], line 12
    print(f"{word:14}: "{char}" appears {count_per_keyword} times.")
          ^
SyntaxError: invalid syntax. Perhaps you forgot a comma?

# Code block without errors
keywords = ["Python", "programming", "exam"]
letter   = "o"

count_total = 0
count_per_keyword = 0

for word in keywords:
    for char in word:
        if char == letter:
            count_per_keyword += 1
    print(f"{word:14}: '{letter}' appears {count_per_keyword} times.")
    count_total += count_per_keyword
    count_per_keyword = 0

print("")
print(f"In total, letter {letter} appears {count_total} times.")
Python        : 'o' appears 1 times.
programming   : 'o' appears 1 times.
exam          : 'o' appears 0 times.

In total, letter o appears 2 times.