pandas-ml-common 0.2.7

The pandas ml common module

This module holds all common extensions and utilities for the pandas ml quant stack. Feel free to study the examples as well.

• easy joining of data frames with multi indexes

from pandas_ml_common import pd, np

df1 = pd.DataFrame({"a": np.random.random(10), "b": np.random.random(10)})
print(df1.inner_join(df1, prefix_left='A', prefix='B', force_multi_index=True).to_markdown())

• access columns with regex

df4 = pd.DataFrame({"a_22_a": np.random.random(1), "b_21_b": np.random.random(1)})
df4._[r'.*\d+_.']

• easy access multi level index

df1.unique_level_columns(0)

['A', 'B']

df1.add_multi_index('Z', axis=1)

• data splitting, sampling and folding (aka cross validation)

from pandas_ml_common import Sampler, XYWeight, random_splitter

df2 = pd.DataFrame({"c": np.random.random(10)})
sampler = Sampler(XYWeight(df1, df2), splitter=random_splitter(0.5))

for batches in sampler.sample_for_training():
    for batch in batches:
        print(batch)

• access to nested numpy arrays in data frame columns (df._.values)

df3 = pd.DataFrame({"a": [[1, 2], [3, 4], [5, 6]]})
df3._.values

array([[1, 2],
       [3, 4],
       [5, 6]])

• dynamic method call providing suitable *args and **kwargs (dependency injection)

from pandas_ml_common import call_callable_dynamic_args

def adder(a, b):
    return a + b

call_callable_dynamic_args(adder, a=12, b=10, c='illegal')

22

• numpy utils

from pandas_ml_common import np_nans

np_nans((3, 3))

array([[nan, nan, nan],
       [nan, nan, nan],
       [nan, nan, nan]])


from pandas_ml_common import temp_seed

with temp_seed(42):
    print(np.random.random(2))

np.random.random(2)


[0.37454012 0.95071431]
array([0.69373278, 0.69790163])

• serialization utils

from pandas_ml_common import serializeb, deserializeb

deserializeb(serializeb(np.array([1, 2, 3])))
array([1, 2, 3])

• re-scalings

from pandas_ml_common import ReScaler

x = np.arange(0, 1, .1)
rescaler = ReScaler((0, 1), (5, -5))

rescaler(x)
array([ 5.,  4.,  3.,  2.,  1.,  0., -1., -2., -3., -4.])