Stationarity and differencing in ML Python - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to difference the time series data once.

ML Python

differenced_series = original_series[1]

Drag options to blanks, or click blank then click option'

A.mean()

B.diff()

C.sum()

D.plot()

Attempts:

3 left

💡 Hint

Common Mistakes

Using .mean() instead of .diff()

Using .sum() which adds values rather than differencing

Trying to plot instead of transforming the data

The .diff() function calculates the difference between consecutive values, which helps to make a time series stationary.

2fill in blank

medium

Complete the code to check if the time series is stationary using the Augmented Dickey-Fuller test.

ML Python

from statsmodels.tsa.stattools import adfuller
result = adfuller(time_series)
print('p-value:', result[1])

Drag options to blanks, or click blank then click option'

A[1]

B[0]

C[4]

D[5]

Attempts:

3 left

💡 Hint

Common Mistakes

Using index 0 which is the test statistic, not p-value

Using index 4 or 5 which are not valid for p-value

The p-value is at index 1 in the result tuple returned by adfuller(). A low p-value indicates stationarity.

3fill in blank

hard

Fix the error in the code to difference the series twice.

ML Python

diff2_series = original_series[1].diff()

Drag options to blanks, or click blank then click option'

A.diff(1)

B.diff(2)

C.diff().diff()

D.diff()

Attempts:

3 left

💡 Hint

Common Mistakes

Using .diff(2) which is not the same as differencing twice

Using .diff(1) which only differences once

To difference twice, you apply diff() two times: .diff().diff(). Using diff(2) calculates the difference between values two steps apart, which is not the same.

4fill in blank

hard

Complete the code to difference the series once, drop NaNs, and print the p-value from the ADF test.

ML Python

differenced = original_series.diff()[1]
result = adfuller(differenced)
print('p-value:', result[2])

Drag options to blanks, or click blank then click option'

A.dropna()

B[1]

C[0]

D.values

Attempts:

3 left

💡 Hint

Common Mistakes

Forgetting .dropna(), causing errors in ADF test

Using [0] (test statistic) instead of [1] (p-value)

Differencing introduces NaN at the start, so chain .dropna() before running adfuller(). The p-value is result[1].

5fill in blank

hard

Fill all three blanks to extract the test statistic and p-value from the ADF test on differenced series, and reference critical values.

ML Python

adf_result = adfuller(differenced_series[1])
test_stat = adf_result[2]
p_value = adf_result[3]
critical_vals = adf_result[4]

Drag options to blanks, or click blank then click option'

A.dropna()

B[0]

C[1]

D[4]

Attempts:

3 left

💡 Hint

Common Mistakes

Omitting .dropna()

Mixing up indices (e.g., using [4] for p-value)

Using wrong tuple positions for stat and p-value

Remove NaNs with .dropna(). The ADF result tuple has test statistic at index [0], p-value at [1], and critical values dictionary at [4].