I’m trying to truncate floating point numbers in my DataFrame to a desired number of decimal places. I’ve found that this can be done using Pandas and NumPy here, but I’ve also seen that it might be possible with polars.Config.set_float_precision.
Below is my current approach, but I think I might be taking extra steps.
import polars as pl
data = {
"name": ["Alice", "Bob", "Charlie"],
"grade": [90.23456, 80.98765, 85.12345],
}
df = pl.DataFrame(data)
df.with_columns(
# Convert to string
pl.col("grade").map_elements(lambda x: f"{x:.5f}", return_dtype=pl.String).alias("formatted_grade")
).with_columns(
# Slice to get my desired decimals
pl.col("formatted_grade").str.slice(0, length = 4)
).with_columns(
# Convert it again to Float
pl.col("formatted_grade").cast(pl.Float64)
)
2
You can use the Polars – Numpy integration like this:
df = df.with_columns(truncated_grade=np.trunc(pl.col("grade") * 10) / 10)
Output:
┌─────────┬──────────┬─────────────────┐
│ name ┆ grade ┆ truncated_grade │
│ --- ┆ --- ┆ --- │
│ str ┆ f64 ┆ f64 │
╞═════════╪══════════╪═════════════════╡
│ Alice ┆ 90.23456 ┆ 90.2 │
│ Bob ┆ 80.98765 ┆ 80.9 │
│ Charlie ┆ 85.12345 ┆ 85.1 │
└─────────┴──────────┴─────────────────┘
Full code:
import numpy as np
import polars as pl
data = {
"name": ["Alice", "Bob", "Charlie"],
"grade": [90.23456, 80.98765, 85.12345],
}
df = pl.DataFrame(data)
df = df.with_columns(truncated_grade=np.trunc(pl.col("grade") * 10) / 10)
print(df)