I have a pandas data frame I want to reorder the index using a specific order.
from random import randint
import pandas as pd
days = ["Tuesday", "Thursday", "Monday", "Wednesday"]
df = pd.DataFrame({"Value": [randint(0, 9) for i in range(len(days)*2)]},
index=[day for day in days for i in range(2)])
myorder = ["Monday", "Tuesday", "Wednesday", "Thursday"]
The specific order is as notified by the list myorder
Use CategoricalIndex + sort_index:
df.index = pd.CategoricalIndex(df.index, categories=myorder, ordered=True)
df = df.sort_index()
print (df)
Value
Monday 3
Monday 3
Tuesday 4
Tuesday 3
Wednesday 3
Wednesday 4
Thursday 5
Thursday 2
This is one method, if you wish to avoid categoricals:
orderdic = dict(zip(myorder, range(len(myorder))))
df = df.assign(order=df.index.to_series().map(orderdic))\
.sort_values('order').drop('order', 1)
# Value
# Monday 6
# Monday 6
# Tuesday 9
# Tuesday 1
# Wednesday 1
# Wednesday 3
# Thursday 0
# Thursday 7
