CODEWITHSUNDEEP
pythonpandasdataframe
Filipe Aleixo
Filipe Aleixo

Reputation: 4242

Pandas logical indexing using multiple conditions

I have a pandas dataframe data_ask_bid indexed by a Datetime stamp for which I want to keep only the rows in the date range: Monday @ 00:00 - Friday @ 21:59. For this, I wrote the following line:

data_ask_bid = data_ask_bid[((0 <= data_ask_bid.index.weekday <= 3) | (data_ask_bid.index.weekday == 4 & data_ask_bid.index.hour < 22))]

Although there seems to be a problem with the logical indexing, as it's throwing the error 'The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()'. Where did I go wrong in the code?

Upvotes: 1

Views: 450

Answers (2)

jezrael
jezrael

Reputation: 863731

I think you can use for check values numpy.in1d:

mask1 = np.in1d(data_ask_bid.index.weekday, [0,1,2,3])
mask2 = data_ask_bid.index.weekday == 4
mask3 = data_ask_bid.index.hour < 22

mask = mask1 | (mask2 & mask3)

data_ask_bid = data_ask_bid[mask]

Sample:

start = pd.to_datetime('2017-02-10 15:00:00')
rng = pd.date_range(start, periods=20, freq='7h')

data_ask_bid = pd.DataFrame({'a': range(20)}, index=rng)  
#print (data_ask_bid)

w = data_ask_bid.index.weekday
mask1 = np.in1d(w, [0,1,2,3])
mask2 = w == 4
mask3 = data_ask_bid.index.hour < 22

mask = mask1 | (mask2 & mask3)
print (mask)
[ True False False False False False False False False  True  True  True
  True  True  True  True  True  True  True  True]

data_ask_bid = data_ask_bid[mask]
print (data_ask_bid)
                      a
2017-02-10 15:00:00   0
2017-02-13 06:00:00   9
2017-02-13 13:00:00  10
2017-02-13 20:00:00  11
2017-02-14 03:00:00  12
2017-02-14 10:00:00  13
2017-02-14 17:00:00  14
2017-02-15 00:00:00  15
2017-02-15 07:00:00  16
2017-02-15 14:00:00  17
2017-02-15 21:00:00  18
2017-02-16 04:00:00  19

Timings:

start = pd.to_datetime('2017-02-10 15:00:00')
N = 1000000
rng = pd.date_range(start, periods=N, freq='H')

data_ask_bid = pd.DataFrame({'a': range(N)}, index=rng)  
print (data_ask_bid)

def jez(data_ask_bid):
    w = data_ask_bid.index.weekday
    mask1 = np.in1d(w, [0,1,2,3])
    mask2 = w == 4
    mask3 = data_ask_bid.index.hour < 22
    data_ask_bid = data_ask_bid[mask1 | (mask2 & mask3)]
    return (data_ask_bid)

print (jez(data_ask_bid))

print (data_ask_bid[(((data_ask_bid.index.weekday >= 0) & (data_ask_bid.index.weekday <= 3)) | ((data_ask_bid.index.weekday == 4) & (data_ask_bid.index.hour < 22)))])

In [273]: %timeit (jez(data_ask_bid))
10 loops, best of 3: 142 ms per loop

In [274]: %timeit (data_ask_bid[(((data_ask_bid.index.weekday >= 0) & (data_ask_bid.index.weekday <= 3)) | ((data_ask_bid.index.weekday == 4) & (data_ask_bid.index.hour < 22)))])
1 loop, best of 3: 267 ms per loop

Upvotes: 1

Filipe Aleixo
Filipe Aleixo

Reputation: 4242

Just found out that Pandas doesn't work with a clause of the type 0 <= data_ask_bid.index.weekday <= 3 so I needed to divide it into 2 separate clauses for it to work:

data_ask_bid = data_ask_bid[(((data_ask_bid.index.weekday >= 0) & (data_ask_bid.index.weekday <= 3)) | ((data_ask_bid.index.weekday == 4) & (data_ask_bid.index.hour < 22)))]

Upvotes: 0

Related Questions