PYTHON

Univariant Variable Analysis - Multiple Plots

# Visualization is the easiest way to have an inference about the overall data and the outliers.
#Plotting Graphs Before treating outliers 
for col in df.describe().columns:
  fig, ax =plt.subplots(1,3, constrained_layout=True)
  fig.set_size_inches(20, 3)
  sns.distplot(df[col], ax=ax[0]).set(title="Distplot")
  sns.histplot(df[col], ax=ax[1]).set(title="Histplot")
  sns.boxplot(df[col], ax=ax[2]).set(title="Boxplot")
  plt.suptitle(f'{col.title()}',weight='bold')
  fig.show()

Univariant Variable Analysis - Multiple Plots

# Defining a function to Notate the percent count of each value on the bars
def annot_percent(axes):
  '''Takes axes as input and labels the percent count of each bar in a countplot'''
  for p in plot.patches:
    total = sum(p.get_height() for p in plot.patches)/100
    percent = round((p.get_height()/total),2)
    x = p.get_x() + p.get_width()/2
    y = p.get_height()
    plot.annotate(f'{percent}%', (x, y), ha='center', va='bottom')
    

# Defining r to autofit the number and size of plots
r = int(len(catagorical_columns)/3 +1)


# Plotting the countplots for each target variable
plt.figure(figsize=(18,r*3))
for n,column in enumerate(catagorical_columns):
  plot = plt.subplot(r,3,n+1)
  sns.countplot(df[column]).margins(y=0.15)
  plt.title(f'{column.title()}',weight='bold')
  plt.tight_layout()
  annot_percent(plot)