CODEWITHSUNDEEP
pythonpdfscipyhistogramdata-fitting
Yujian
Yujian

Reputation: 169

Scaling of fitted pdf for a histogram

I'm working on some time series data and I tried to fit the data with a gamma distribution. But the problem is that the magnitude of fitted pdf is much lower than that of the histogram. Here are my code and plot. What is wrong with the plot?

# the data
data = contents[0][1:]

# normalized histogram
weights = np.ones_like(data)/len(data)
plt.hist(data, bins = 20, color = 'w', edgecolor = 'black', alpha = 0.5, weights = weights)

# fit with gamma distribution and plot the pdf
dist = getattr(scipy.stats, 'gamma')
param = dist.fit(data)
x = np.linspace(min(data), max(data), 100)
pdf_fit = dist.pdf(x, *param[:-2], loc = param[-2], scale = param[-1])
plt.plot(x, pdf_fit/sum(pdf_fit), label = 'Gamma')
plt.legend(loc = 'upper right')
plt.show()

enter image description here

Upvotes: 0

Views: 691

Answers (1)

Warren Weckesser
Warren Weckesser

Reputation: 114821

In your call to plt.hist(), instead of using weights=np.ones_like(data)/len(data), use the argument normed=True:

plt.hist(data, bins = 20, color = 'w', edgecolor = 'black', alpha = 0.5, normed = True)

Upvotes: 1

Related Questions