Histogram
- Machine Learning course
- Beginner
- Lesson 105/112
- 5 min
A histogram shows data distribution with bars.
Histogram
A histogram shows the distribution of a single variable: the value range is split into bins (bins) and the number of examples in each bin is drawn as a bar. Skew, multiple peaks and outliers show up here.
fig, ax = plt.subplots()
ax.hist(x, bins=20, edgecolor="black")
sayi, sinir = np.histogram(x, bins=20) # same counts, no drawingnp.histogram gives the same bin counts and edges without drawing.
Bin count and comparison
Few bins hide detail, many bins produce noise. To compare classes, histograms are drawn semi-transparent on the same axis:
for k in [0, 1]:
ax.hist(X[y == k, 0], bins=20, alpha=0.5, label=f"sınıf {k}")
ax.legend()If the two classes' histograms overlap a lot, that feature is weak at separating the classes.
Practice Count the distribution
Draw a 4-bin histogram of x and save it. Get the same counts with np.histogram(x, bins=4); print the counts and the edges on separate lines.
Starter code
import numpy as np
import matplotlib.pyplot as plt
x = np.array([1, 2, 2, 3, 3, 3, 4, 4, 5, 9])
# buraya yazSample solution
import numpy as np
import matplotlib.pyplot as plt
x = np.array([1, 2, 2, 3, 3, 3, 4, 4, 5, 9])
fig, ax = plt.subplots()
ax.hist(x, bins=4)
fig.savefig("hist.png")
sayi, sinir = np.histogram(x, bins=4)
print(sayi)
print(sinir)Expected output
[3 5 1 1] [1. 3. 5. 7. 9.]
Questions and answers
Review the lesson questions together with their answers. To test yourself, solve the lesson interactively.
1 How many bin edges does <code>np.histogram(x, bins=10)</code> return?
- 9
- 20
- 11 Correct answer
- 10
Explanation: The starts and ends of 10 bins: 11 edges; the count array has 10 elements.
2 Show the legend.
ax.____()ax.legend()Explanation: legend draws the key for plots that were given a label.
3 Match the concepts.
| Term | Meaning |
|---|---|
bins |
Number of bins |
np.histogram |
Counting without drawing |
label |
Legend text |
overlapping histogram |
Weak separating feature |