Pandas DataFrame.loc returns Empty DataFrame

user_7729934 0
Viewed 5792

Currently trying to work through some tutorials. Right now I have a pandas Dataframe with stock data. The stock data was taken by reading a csv file, then using df.set_index('timestamp', inplace = True) I set the index to use 'timestamp'. The head is seen here:

 timestamp   open   high    low   close  adjusted_close   volume  dividend_amount  split_coefficient
2018-09-11  74.95  75.69  74.76  75.64           75.64  2225700              0.0                1.0
2018-09-10  75.10  75.21  74.84  74.91           74.91  1774400              0.0                1.0
2018-09-07  75.20  75.20  74.72  75.01           75.01  1804800              0.0                1.0
2018-09-06  74.95  75.55  74.62  75.24           75.24  3058300              0.0                1.0
2018-09-05  75.03  75.42  74.80  74.95           74.95  2418400              0.0                1.0

When I try df.loc['2018-09-05'], it returns the correct row to me. However, when I try to select a range, such as df.loc['2018-09-05':'2018-09-11'], I get an Empty DataFrame returned as so:

Empty DataFrame
Columns: [open, high, low, close, adjusted_close, volume, dividend_amount, 
split_coefficient]
Index: []

Was wondering if anyone could provide any insight as to why this is happening? I would have expected a return of the information between the two dates instead of the empty dataframe. Thanks!

pandas dataframe python
5 Answers

sort_index() before slicing would work:

df = df.sort_index()
df['2018-09-05':'2018-09-11']

or

df = df.sort_index()
df.loc['2018-09-05':'2018-09-11']

FYI: If you want to make sure that index is DatetimeIndex before making 'timestamp' as index:

df.timestamp = pd.to_datetime(df.timestamp)
df = df.set_index('timestamp')

or after the fact:

df.index = pd.to_datetime(df.index)
user_44723310

Because you need to insert your index correctly from up to bottom. Try this for specific indexing:

df.loc['2018-09-11':'2018-09-05']

But you can use sort() to sort your index.
You can also use the ID for this.

user_103373540

If the timestamp column is Pandas Timestamp, You can do something like this.

mask = (df['timestamp'] >= pd.Timestamp('2018-09-05')) & (df['timestamp'] <= pd.Timestamp('2018-09-11'))

df.loc[mask, :]
user_78968490

You just needed to reverse dates from newer to older and it will work. 

import pandas as pd
df = pd.read_csv('some.csv', delimiter=";", index_col='timestamp')
range = df.loc['2018-09-11':'2018-09-05']
print(range)
user_64999480

It works if you change the date format

df.loc['05/09/2018': '11/09/2018']
user_18892970
Related