# First Python Data Example
# Module 1 - First Python data example
# Dataset: /data/wage_sample.csv

# First Python Data Example
#
# Use Python to describe a wage and education relationship. Keep the interpretation descriptive.

# %% Cell 2
import pandas as pd

df = pd.read_csv("wage_sample.csv")
print(df[["wage", "education", "experience"]].describe())

# %% Cell 3
correlation = df["wage"].corr(df["education"])
print("Correlation:", round(correlation, 3))

# %% Cell 4
grouped = df.groupby("education")["wage"].mean()
print(grouped)

# Careful interpretation
#
# The output can describe an association in this sample. It should not be treated as proof that education causes wages to rise without stronger assumptions or research design.