Peak finding (find_peaks) in SciPy

Introduction

We use peak finding to locate the highest points in data. This helps us understand important events or changes.

Finding the highest sales days in a month from daily sales data.

Detecting heartbeats from a noisy ECG signal.

Identifying the loudest moments in an audio recording.

Spotting the tallest waves in ocean wave height data.

Syntax

SciPy

scipy.signal.find_peaks(x, height=None, threshold=None, distance=None, prominence=None, width=None)

x is the data array where you want to find peaks.

You can add conditions like minimum height or distance between peaks to control the results.

Examples

Find all peaks in the data without any conditions.

SciPy

from scipy.signal import find_peaks

peaks, _ = find_peaks(data)

Find peaks that have a height of at least 5.

SciPy

peaks, _ = find_peaks(data, height=5)

Find peaks that are at least 3 points apart.

SciPy

peaks, _ = find_peaks(data, distance=3)

Sample Program

This code finds peaks in the array x where the peak height is at least 2. It prints the data, the positions of the peaks, and their heights.

SciPy

import numpy as np
from scipy.signal import find_peaks

# Create sample data with peaks
x = np.array([0, 2, 1, 3, 7, 1, 2, 6, 0, 1])

# Find peaks
peaks, properties = find_peaks(x, height=2)

print("Data:", x)
print("Peak indices:", peaks)
print("Peak heights:", properties['peak_heights'])

Important Notes

Peaks are points higher than their neighbors.

You can use prominence to find peaks that stand out more clearly.

Distance helps avoid detecting peaks that are too close together.

Summary

Use find_peaks to locate high points in data.

You can set conditions like height and distance to control which peaks are found.

This helps analyze patterns and important events in data.

Practice

1. What does the find_peaks function in scipy do?

easy

A. It finds local maxima (peaks) in a 1D array of data.

B. It sorts the data in ascending order.

C. It calculates the average value of the data.

D. It removes all negative values from the data.

Solution

1. Step 1: Understand the purpose of find_peaks

   The function is designed to identify points in data where values are higher than their neighbors, called peaks.

2. Step 2: Compare options with function behavior

   Only It finds local maxima (peaks) in a 1D array of data. correctly describes this behavior; others describe unrelated operations.

3. Final Answer:

   It finds local maxima (peaks) in a 1D array of data. -> Option A

4. Quick Check:

   find_peaks locates peaks = D [OK]

Hint: Remember: find_peaks finds highs, not sorting or averaging [OK]

Common Mistakes:

• Confusing peak finding with sorting data
• Thinking it calculates statistics like mean
• Assuming it filters values instead of locating peaks

2. Which of the following is the correct way to import and use find_peaks from scipy.signal?

easy

A. import find_peaks from scipy.signal
peaks = find_peaks(data)

B. from scipy.signal import find_peaks
peaks = find_peaks(data)

C. from scipy import find_peaks
peaks = find_peaks(data)

D. import scipy.signal.find_peaks
peaks = find_peaks(data)

Solution

1. Step 1: Recall correct import syntax in Python

   Python uses 'from module import function' to import specific functions.

2. Step 2: Match syntax with options

   from scipy.signal import find_peaks peaks = find_peaks(data) uses correct syntax. Options A, C, and D have invalid import statements.

3. Final Answer:

   from scipy.signal import find_peaks\npeaks = find_peaks(data) -> Option B

4. Quick Check:

   Correct import syntax = B [OK]

Hint: Use 'from module import function' for specific imports [OK]

Common Mistakes:

• Using 'import function from module' which is invalid
• Trying to import from wrong module path
• Using dot notation in import statement incorrectly

3. Given the code below, what will be the output of peaks?

import numpy as np
from scipy.signal import find_peaks

data = np.array([0, 2, 1, 3, 0, 1, 2, 1])
peaks, _ = find_peaks(data, height=2)
print(peaks)

medium

A. [3]

B. [1 3]

C. [1 3 6]

D. [2 5]

Solution

1. Step 1: Identify peaks in the data array

   Data values: [0, 2, 1, 3, 0, 1, 2, 1]. Peaks are points higher than neighbors: indices 1 (2), 3 (3), and 6 (2).

2. Step 2: Apply height=2 filter

   Peaks with height >= 2 are kept. Index 1 (2 >= 2), 3 (3 >= 2), 6 (2 >= 2), so indices 1, 3, 6 remain.

3. Final Answer:

   [1 3 6] -> Option C

4. Quick Check:

   Peaks with height >= 2 = [1 3 6] [OK]

Hint: Check peak values against height filter carefully [OK]

Common Mistakes:

• Thinking height filter requires strictly greater than threshold (it's >=)
• Misidentifying peak positions
• Ignoring the height parameter

4. The following code is intended to find peaks at least 3 units apart, but it returns more peaks than expected. What is the error?

import numpy as np
from scipy.signal import find_peaks

data = np.array([1, 3, 2, 4, 3, 5, 2])
peaks = find_peaks(data, 3)
print(peaks)

medium

A. The distance parameter is not applied because it should be passed as a keyword argument.

B. The distance parameter is applied correctly; the issue is with the data array.

C. The code is missing parentheses around the print argument.

D. The find_peaks function does not support the distance parameter.

Solution

1. Step 1: Check how distance parameter is passed

   In find_peaks, distance must be passed as a keyword argument, e.g., distance=3.

2. Step 2: Analyze the code

   The code passes 3 positionally after data, which sets height=3 instead of distance=3. Thus distance is not applied, returning all local peaks [1,3,5] instead of filtered [1,5].

3. Step 3: Verify output type

   find_peaks returns a tuple (indices, properties). Printing peaks directly prints the tuple, which may confuse.

4. Final Answer:

   The distance parameter is not applied because it should be passed as a keyword argument. -> Option A

5. Quick Check:

   distance must be keyword argument = C [OK]

Hint: Always pass distance as keyword argument: distance=number [OK]

Common Mistakes:

• Passing distance as positional argument
• Expecting find_peaks to return only indices
• Misunderstanding output format of find_peaks

5. You have noisy sensor data and want to find peaks that are at least 5 units high and at least 10 samples apart. Which find_peaks call correctly applies these conditions?

import numpy as np
from scipy.signal import find_peaks

sensor_data = np.random.rand(100) * 10

# Which call is correct?

hard

A. peaks = find_peaks(sensor_data, height=(5, None), distance=10)

B. peaks = find_peaks(sensor_data, height=5, distance=10)

C. peaks, _ = find_peaks(sensor_data, min_height=5, min_distance=10)

D. peaks, _ = find_peaks(sensor_data, height=5, distance=10)

Solution

1. Step 1: Understand parameter usage in find_peaks

   Height is specified with 'height=5' to find peaks at least 5 units high. Distance is specified with 'distance=10' to ensure peaks are at least 10 samples apart.

2. Step 2: Evaluate each option

   peaks, _ = find_peaks(sensor_data, height=5, distance=10) correctly uses height and distance parameters and unpacks the tuple output. peaks = find_peaks(sensor_data, height=5, distance=10) misses unpacking. peaks, _ = find_peaks(sensor_data, min_height=5, min_distance=10) uses invalid parameter names. peaks = find_peaks(sensor_data, height=(5, None), distance=10) uses a tuple for height which is valid but unnecessary here.

3. Final Answer:

   peaks, _ = find_peaks(sensor_data, height=5, distance=10) -> Option D

4. Quick Check:

   Use height and distance keywords, unpack output = A [OK]

Hint: Use height= and distance= with unpacking for correct results [OK]

Common Mistakes:

• Using wrong parameter names like min_height
• Not unpacking the tuple returned by find_peaks
• Passing height as a tuple unnecessarily