Python with AI Cheat Sheet

Writing Python for data science is faster when AI handles the syntax. This cheat sheet covers prompting principles that produce reliable code, plus ready-to-run snippets for the packages and data structures you use every day: pandas, NumPy, lists, dictionaries, strings, and file paths.

Principles of good Python prompting

When using AI to generate Python, vague prompts produce fragile code. Clear prompts produce reliable code.

1. State the task clearly

Bad:

Do something with this DataFrame.

Good:

From sales_df, calculate the mean of the revenue column and store it in a variable called mean_revenue.

2. Specify the package

Bad:

Load this file.

Good:

Use pandas to read data/sales.csv into a DataFrame called sales_df.

3. Define inputs and outputs

Bad:

Clean the data.

Good:

In sales_df, fill missing values in revenue with 0 and return the updated DataFrame.

4. Add constraints

Bad:

Filter the rows.

Good:

Filter sales_df to keep rows where revenue > 1000 using df.query (avoid .loc).

5. Ask for runnable code

Bad:

Show me how to calculate a mean.

Good:

Write complete, runnable Python code (including imports) to calculate the mean of a NumPy array called values.

Core data science packages

A package is a collection of Python modules that provide additional functionality. An alias is an easy-to-type name assigned to a package when it is imported.

Import common packages

Prompt:

Import numpy, pandas, and pathlib using standard aliases.

Generated Python:

import numpy as np
import pandas as pd
from pathlib import Path

Working with DataFrames (pandas)

A DataFrame is a two-dimensional table of labeled rows and columns, provided by the pandas package. A Series is a single column of a DataFrame.

Read a CSV file

Prompt:

Use pandas to read data/sales.csv into a DataFrame called sales_df. Parse a date column as dates.

Generated Python:

import pandas as pd
sales_df = pd.read_csv("data/sales.csv", parse_dates=["date"])

Quickly inspect a DataFrame

Prompt:

Show the first 5 rows, column info, and summary statistics for numeric columns in sales_df.

Generated Python:

print(sales_df.head())
print(sales_df.info())
print(sales_df.describe())

Filter rows

Prompt:

Filter sales_df to keep rows where revenue >= 1000 and region == "EMEA" using df.query.

Generated Python:

filtered_sales_df = sales_df.query("revenue >= 1000 and region == 'EMEA'")

Create new columns

Prompt:

From sales_df, create a new column profit = revenue - cost.

Generated Python:

sales_df["profit"] = sales_df["revenue"] - sales_df["cost"]

Handle missing values

Prompt:

In sales_df, count missing values per column and fill missing revenue with 0.

Generated Python:

print(sales_df.isna().sum())
sales_df["revenue"] = sales_df["revenue"].fillna(0)

Aggregate by group

Prompt:

Group sales_df by region and calculate total and average revenue.

Generated Python:

region_summary_df = (
    sales_df.groupby("region", as_index=False)
    .agg(
        total_revenue=("revenue", "sum"),
        avg_revenue=("revenue", "mean"),
    )
)

Sort and select top rows

Prompt:

Show the top 5 rows of sales_df sorted by revenue descending.

Generated Python:

top_sales_df = sales_df.sort_values("revenue", ascending=False).head(5)

Working with NumPy arrays

A NumPy array is a fast container for numerical data. NumPy is vectorized, so you can perform operations on rows or columns without writing explicit loops.

Create arrays

Prompt:

Create a NumPy array called values from [1, 2, 3, 4, 5].

Generated Python:

import numpy as np
values = np.array([1, 2, 3, 4, 5])

Filter with boolean indexing

Prompt:

From values, keep only numbers greater than 3.

Generated Python:

values_over_3 = values[values > 3]

Calculate summary statistics

Prompt:

Using NumPy, calculate the mean, standard deviation, maximum, and 25th percentile of values.

Generated Python:

mean_value = np.mean(values)
std_value = np.std(values)
max_value = np.max(values)
q25 = np.quantile(values, 0.25)

Working with lists

A list is an ordered, mutable sequence that can hold elements of any type.

Create and transform a list

Prompt:

Given numbers = [1, 2, 3, 4], create a new list containing the squares.

Generated Python:

numbers = [1, 2, 3, 4]
squares = [n ** 2 for n in numbers]

Filter a list

Prompt:

Given numbers = [1, 2, 3, 4, 5, 6], create a new list containing only the even numbers.

Generated Python:

numbers = [1, 2, 3, 4, 5, 6]
even_numbers = [n for n in numbers if n % 2 == 0]

Working with dictionaries

A dictionary stores data as key-value pairs.

Create and access dictionary values

Prompt:

Create a dictionary mapping model names to accuracy scores and retrieve the value for "xgboost" safely.

Generated Python:

model_accuracy = {"baseline": 0.78, "xgboost": 0.85, "nn": 0.83}
xgboost_score = model_accuracy.get("xgboost", None)

Iterate over dictionary items

Prompt:

Given a dictionary model_accuracy, print each model name and accuracy in the format model: accuracy.

Generated Python:

model_accuracy = {"baseline": 0.78, "xgboost": 0.85, "nn": 0.83}
for model_name, accuracy in model_accuracy.items():
    print(f"{model_name}: {accuracy}")

Working with text

A string is a sequence of characters. An f-string is a formatted string that embeds variables using {}.

Format text with f-strings

Prompt:

Print the mean revenue with 2 decimal places using an f-string.

Generated Python:

mean_revenue = sales_df["revenue"].mean()
print(f"Mean revenue: {mean_revenue:.2f}")

Clean text (strip and lowercase)

Prompt:

Given a string raw_text = "  Data Science  ", remove leading/trailing spaces and convert it to lowercase.

Generated Python:

raw_text = "  Data Science  "
clean_text = raw_text.strip().lower()
print(clean_text)

Working with file paths

Use the pathlib package for modern file handling.

Construct file paths

Prompt:

Use pathlib to construct a path to data/sales.csv.

Generated Python:

from pathlib import Path
project_root = Path(".")
sales_path = project_root / "data" / "sales.csv"

Read a text file

Prompt:

Using pathlib, read the contents of notes.txt into a string called notes_text.

Generated Python:

from pathlib import Path
notes_path = Path("notes.txt")
notes_text = notes_path.read_text(encoding="utf-8")