Back to exercises

Exercise 7

Part 1 — APIs and Interpolation

In this first part, you will be using the free weather information API mentioned in the lecture. Download the necessary CSV file from here: https://learn-data-science-site.vercel.app/locations.csv

  • Open and inspect this file with pandas

  • Plot the relative locations of the cities based on longitude and latitude

    • Plot Garching (48.26, 11.67) in a different color than the other cities. Note: please do not add Garching to the DataFrame.
    • You may negate the Earth's curvature for this plot due to the relative closeness of the cities
    • Add the name of each city to the plot. Hint: look up plt.text()

  • Check out the API documentation of this site we briefly looked at in class: https://open-meteo.com/en/docs

    • Write a python function using requests that takes in the latitude and longitude of a location and directly returns the current temperature as a float. Build in a small delay (at least 1s) before each request to avoid rate limiting.
      • Recall: use Python's built-in time module for this delay.
    • Apply this function to your DataFrame to get a new column corresponding the current temperature at each location.

  • Use the CloughTocher2DInterpolator from SciPy to create an interpolator which can estimate the temperature anywhere between the given cities.

    • Use numpy's linspace and meshgrid to generate a coordinate grid (X and Y variables below)
    • Get the interpolated temperature at each of these X,Y coordinates (Z below)
    • Create a mesh plot with a color bar using the following code:
    plt.pcolormesh(Y, X, Z)
plt.colorbar()
plt.show()
    • Plot your cities on top of this mesh grid plot (likely you can copy the code from above for easy overplotting)

  • Finally, use your interpolator to estimate the current temperature in Garching. Does this look close to the current temperature given on your phone?

Part 2 - Fast Fourier Transform

sample_rate, audio_data = scipy.io.wavfile.read(wav_filepath)
print(f"Shape of stereo audio: {audio_data.shape}")
left_channel, right_channel = audio_data[:,0], audio_data[:,1]

magnitude = np.abs(scipy.fft.rfft(left))
frequency = scipy.fft.rfftfreq(len(left), 1 / sample_rate)
  • Plot frequency vs magnitude:
    • Find a suitable maximum cutoff for the frequency of the plot, so you can more clearly see the important part of the frequency spac
      • Hint: use plt.xlim()
    • How many peaks do you see more or less?
  • What frequency is heard most in the piece of music? I.e you must calculate the frequency that has the highest peak in this plot above.
    • Use this chart to find the musical note closest to this frequency: https://muted.io/note-frequencies
    • E.g. if you found the frequency to be 46Hz, the closest musical note would be F#1.
    • Whatever note you find is the secret word ✨

Submit the answer: