I have a custom function previously working pretty fine in Pandas, but somehow it won't work in the polars map_elements() function when I am trying to migrate to polars.
below is my code.
df = df.sort("time", descending=False)
latest_day = df[-1,'time']
def fetch_price(self, date): # the custom function
if date <= latest_day:
return df[date,self.name]
else:
return None
def get_rate(self, direction, margin, test_cycle):
# retrieve relevant data from df
self.df = (df[['time', self.name]].drop_nulls())[
test_cycle * -1:] # test_cycle*-1. test_cycle大于life_cycle。
# apply next_trade_day to 'time' column, and create a new column 'Maturity_Date'
self.df = self.df.with_columns(
(pl.col('time').map_elements(lambda x: next_trade_day(x, self.life_cycle))).alias('Maturity_Date'))
# check if the generated dates are in the original time series
self.df = self.df.with_columns(pl.col('Maturity_Date').is_in(df['time']).cast(bool).alias('Data_Accessibility'))
# maturity price on the day of maturity
# apply fetch_price to 'Maturity_Date' column if 'Data_Accessibility' is True, and create a new column 'Maturity_Price'
# self.df = self.df.with_columns((pl.col('Maturity_Date').map_elements(lambda x: self.fetch_price(x))).alias('Maturity_Price'))
self.df = self.df.with_columns(pl.when(pl.col('Maturity_Date') <= pl.col('time').last()).then(pl.col(self.name)).alias('Maturity_Price'))
print(self.df)
CU = future('CU8888.SHF', 33)
CU.get_rate(1, 0.05, 8000)
This is the output of print (self.df) :
┌────────────┬────────────┬───────────────┬────────────────────┐
│ time ┆ CU8888.SHF ┆ Maturity_Date ┆ Data_Accessibility │
│ --- ┆ --- ┆ --- ┆ --- │
│ date ┆ f64 ┆ date ┆ bool │
╞════════════╪════════════╪═══════════════╪════════════════════╡
│ 2004-01-02 ┆ 23275.0 ┆ 2004-02-04 ┆ true │
│ 2004-01-02 ┆ 23275.0 ┆ 2004-02-04 ┆ true │
│ 2004-01-02 ┆ 23275.0 ┆ 2004-02-04 ┆ true │
│ 2004-01-02 ┆ 23275.0 ┆ 2004-02-04 ┆ true │
│ … ┆ … ┆ … ┆ … │
│ 2023-03-23 ┆ 68274.0 ┆ 2023-04-25 ┆ false │
│ 2023-03-24 ┆ 69670.0 ┆ 2023-04-26 ┆ false │
│ 2023-03-27 ┆ 69161.0 ┆ 2023-05-04 ┆ false │
│ 2023-03-28 ┆ 69244.0 ┆ 2023-05-04 ┆ false │
└────────────┴────────────┴───────────────┴────────────────────┘
The main issue here was you had a datetime index in pandas and could use df.loc[date, ...] to do a "row lookup".
There is no .loc or indexes in polars, however, the .apply + .loc approach being used is essentially mimicking the functionality of a left join.
df = pl.from_repr("""
┌────────────┬────────────┬───────────────┬────────────────────┐
│ time ┆ CU8888.SHF ┆ Maturity_Date ┆ Data_Accessibility │
│ --- ┆ --- ┆ --- ┆ --- │
│ date ┆ f64 ┆ date ┆ bool │
╞════════════╪════════════╪═══════════════╪════════════════════╡
│ 2004-01-02 ┆ 23275.0 ┆ 2004-02-04 ┆ true │
│ 2004-01-02 ┆ 23275.0 ┆ 2004-02-04 ┆ true │
│ 2004-01-02 ┆ 23275.0 ┆ 2004-02-04 ┆ true │
│ 2004-01-02 ┆ 23275.0 ┆ 2004-02-04 ┆ true │
│ 2004-02-04 ┆ 99999.0 ┆ 2004-05-10 ┆ false │
│ 2023-03-23 ┆ 68274.0 ┆ 2023-04-25 ┆ false │
│ 2023-03-24 ┆ 69670.0 ┆ 2023-04-26 ┆ false │
│ 2023-03-27 ┆ 69161.0 ┆ 2023-05-04 ┆ false │
│ 2023-03-28 ┆ 69244.0 ┆ 2023-05-04 ┆ false │
└────────────┴────────────┴───────────────┴────────────────────┘
""")
df.join(
df.select("time", Maturity_Price = "CU8888.SHF"),
left_on="Maturity_Date",
right_on="time",
how="left"
)
shape: (9, 5)
┌────────────┬────────────┬───────────────┬────────────────────┬────────────────┐
│ time ┆ CU8888.SHF ┆ Maturity_Date ┆ Data_Accessibility ┆ Maturity_Price │
│ --- ┆ --- ┆ --- ┆ --- ┆ --- │
│ date ┆ f64 ┆ date ┆ bool ┆ f64 │
╞════════════╪════════════╪═══════════════╪════════════════════╪════════════════╡
│ 2004-01-02 ┆ 23275.0 ┆ 2004-02-04 ┆ true ┆ 99999.0 │
│ 2004-01-02 ┆ 23275.0 ┆ 2004-02-04 ┆ true ┆ 99999.0 │
│ 2004-01-02 ┆ 23275.0 ┆ 2004-02-04 ┆ true ┆ 99999.0 │
│ 2004-01-02 ┆ 23275.0 ┆ 2004-02-04 ┆ true ┆ 99999.0 │
│ … ┆ … ┆ … ┆ … ┆ … │
│ 2023-03-23 ┆ 68274.0 ┆ 2023-04-25 ┆ false ┆ null │
│ 2023-03-24 ┆ 69670.0 ┆ 2023-04-26 ┆ false ┆ null │
│ 2023-03-27 ┆ 69161.0 ┆ 2023-05-04 ┆ false ┆ null │
│ 2023-03-28 ┆ 69244.0 ┆ 2023-05-04 ┆ false ┆ null │
└────────────┴────────────┴───────────────┴────────────────────┴────────────────┘