PYTHON

using pandas stack and subset to return a dataframe object of highly correated pairs

# save absolute value of correlation matrix as a data frame
# converts all values to absolute value
# stacks the row:column pairs into a multindex
# reset the index to set the multindex to seperate columns
# sort values. 0 is the column automatically generated by the stacking

df= old_df.corr().abs().stack().reset_index().sort_values(0, ascending=False)

# zip the variable name columns (Which were only named level_0 and level_1 by default) 
# in a new column named "pairs".
df['pairs'] = list(zip(df.level_0, df.level_1))

# set index to pairs
df.set_index(['pairs'], inplace = True)

#d rop level columns
df.drop(columns=['level_1', 'level_0'], inplace = True)

# rename correlation column as cc rather than 0
df.columns = ['cc']

# drop duplicates. This could be dangerous if you have variables perfectly correlated with variables other than themselves.
# for the sake of exercise, kept it in.
df.drop_duplicates(inplace=True)

# subset to include desired range
df[(df.cc>.75) & (df.cc <1)]