4.4.3 Heatmap

The heatmap is a way of representing the data in a 2-dimensional form. The data values are represented as colors in the graph. The goal of the heatmap is to provide a colored visual summary of information.

1. Heatmap and Annotation

We will create a random dataset and a subplot. The left one is the default mode heatmap, the right one shows the annotation.

# Create a random dataset 
data = np.random.rand(5, 10)

f,axes = plt.subplots(1,2, figsize = (20,6))
sns.heatmap(data, ax=axes[0])
sns.heatmap(data, annot=True, ax=axes[1])

Simple Heatmap

2. Horizontal Bar Heatmap

Meanwhile, we can change the color bar to horizontal.

sns.heatmap(data, annot=True, 
        cbar_kws={'label': 'Horizontal Colorbar', 'orientation': 'horizontal'})

Horizontal Colorbar Heatmap

In General, the darker color represents the larger value, the lighter color represent the smaller value. The above example violates the principle. So, we need to modify the color set bycmapparameter

Also, with the annotation display, we can read the information directly. To make the graph as clean and simple as possible, we can remove the color bar.

f,axes = plt.subplots(1,2, figsize = (20,6))
sns.heatmap(data,cmap="YlGnBu",ax=axes[0])
sns.heatmap(data,cmap="YlGnBu",annot=True, cbar=False,ax=axes[1])

3. Categorical Heatmap (in Triangle)

# create a simple dataset,giving the column names.
d = pd.DataFrame(data,
                 columns=['A','B','C','D','E','F','G','H','I','J'])
                 
# Compute the correlation matrix
corr = d.corr()

# Generate a mask for the upper triangle
mask = np.triu(np.ones_like(corr, dtype=np.bool))

# Generate a custom diverging colormap
cmap = sns.diverging_palette(220, 10, as_cmap=True)

# Draw the heatmap with the mask and correct aspect ratio
sns.heatmap(corr,mask = mask,cmap=cmap, vmax=.3, center=0,
            square=True, linewidths=.5, cbar_kws={"shrink": .5})

Advanced Heatmap