How to wrap ta-lib function as a Polars expression

Question

I am trying to call some TA-lib(https://github.com/mrjbq7/ta-lib) functions through Polars so that the multiple stocks' technical indicators could possibly be calculated through Polars' parallel computing framework.

Here is the sample code

import talib
import polars as pl
import yfinance as yf

tesla = yf.Ticker('TSLA')
tesla_data = tesla.history(period="1Y")
tesla_data["Date"]=tesla_data.index
pl_df = pl.from_pandas(tesla_data[["Date", "Open", "High", "Low", "Close", "Volume"]])

# Method 1. Using ta-lib as a direct function call.
mv_kama = talib.KAMA(pl_df["Close"], 30)

# Method 2. Using ta-lib as Polars expression
def kama30() -> pl.Expr:
    return talib.KAMA(pl.col("Close"), 30)

pl_df2 = pl_df.select([
    pl.col("Close"),
    kama30()
])

The method 2 code snippet however failed to run and the error message was:

TypeError                                 Traceback (most recent call last)
Input In [5], in <cell line: 17>()
     14 def kama30() -> pl.Expr:
     15     return talib.KAMA(pl.col("Close"), 30)
     17 pl_df2 = pl_df.select([
     18     pl.col("Close"),
---> 19     kama30()
     20 ])

Input In [5], in kama30()
     14 def kama30() -> pl.Expr:
---> 15     return talib.KAMA(pl.col("Close"), 30)

File C:\ProgramData\Anaconda3\envs\Charm3.9\lib\site-packages\talib\__init__.py:64, in _wrapper.<locals>.wrapper(*args, **kwds)
     61     _args = args
     62     _kwds = kwds
---> 64 result = func(*_args, **_kwds)
     66 # check to see if we got a streaming result
     67 first_result = result[0] if isinstance(result, tuple) else result

TypeError: Argument 'real' has incorrect type (expected numpy.ndarray, got Expr)

Appreciate if someone could advise how this could be done properly.

Thanks!

I have no experience with polars, but I'm not sure if I'm specifying the wrong columns. I was able to get the values with the following. `return talib.KAMA(pl_df['Close'], 30)` — r-beginners, Apr 13 '22 at 09:23
Thank you for the advice. I followed your suggestion and changed the Polars expression to (even though I am not quite sure we should use pl_df directly) ```def kama30() -> pl.Expr: return talib.KAMA(pl_df.col("Close"), 30)```, I still got the error message: ```AttributeError: col``` — Peascod, Apr 13 '22 at 09:44

score 0 · Answer 1 · answered Apr 14 '22 at 08:53

There is no need to define a Polars function expression, all you need is to use the map function. Here is the working code

import talib
import polars as pl
import yfinance as yf

tesla = yf.Ticker('TSLA')
tesla_data = tesla.history(period="1Y")
tesla_data["Date"]=tesla_data.index
pl_df = pl.from_pandas(tesla_data[["Date", "Open", "High", "Low", "Close", "Volume"]])


pl_df2 = pl_df.select([
    pl.col("Close"),
    pl.col("Close").map(lambda x: talib.KAMA(x, 30)).alias("KAMA30")
])

score 0 · Answer 2 · answered Jan 14 '23 at 14:11

Change line 15 from:

return talib.KAMA(pl.col("Close"), 30)

to:

return talib.KAMA(pl_df["Close"], 30)

To complete you can also add the columname, from:

pl_df2 = pl_df.select([
    pl.col("Close"),
    kama30()
])

to:

pl_df2 = pl_df.select([
    pl.col("Close"),
    pl.Series(kama30()).alias('kama30')
])

So the code will be:

import talib
import polars as pl
import yfinance as yf

tesla = yf.Ticker('TSLA')
tesla_data = tesla.history(period="1Y")
tesla_data["Date"]=tesla_data.index
pl_df = pl.from_pandas(tesla_data[["Date", "Open", "High", "Low", "Close", "Volume"]])

# Method 1. Using ta-lib as a direct function call.
mv_kama = pl.Series(talib.KAMA(pl_df["Close"], 30)).alias('kama30')

# Method 2. Using ta-lib as Polars expression
def kama30() -> pl.Expr:
    return talib.KAMA(pl_df["Close"], 30)

pl_df2 = pl_df.select([
    pl.col("Close"),
    pl.Series(kama30()).alias('kama30')
])

print(mv_kama)
print(pl_df2)

In order to keep all dataframe columns ask for pl.all() too:

import talib
import polars as pl
import yfinance as yf

tesla = yf.Ticker('TSLA')
tesla_data = tesla.history(period="1Y")
tesla_data["Date"]=tesla_data.index
pl_df = pl.from_pandas(tesla_data[["Date", "Open", "High", "Low", "Close", "Volume"]])

# Method 1. Using ta-lib as a direct function call.
# mv_kama = pl.Series(talib.KAMA(pl_df["Close"], 30)).alias('kama30')

# Method 2. Using ta-lib as Polars expression
def kama30() -> pl.Expr:
    return talib.KAMA(pl_df["Close"], 30)

pl_df2 = pl_df.select([
    pl.all(),
    # pl.col("Close"),
    pl.Series(kama30()).alias('kama30')
])

# print(mv_kama)
print(pl_df2)

How to wrap ta-lib function as a Polars expression

2 Answers2