4.3 pandas DataFrames

Big ideas

• import pandas as pd. A DataFrame is a table with named columns; a Series is one column. pd.read_csv("file.csv") loads data; pd.DataFrame({...}) builds one from lists.
• First look: df.head() (5 rows), df.shape, df.columns, df.info(), df.describe().
• Select: df["col"] a column, df[["a", "b"]] several, df.loc[row_label, "col"] by label, df.iloc[0] by position. Filter with a mask: df[df["score"] > 80].
• Compute: df["col"].mean(), .max(), .value_counts(), df.sort_values("col", ascending=False), new columns df["pct"] = df["score"] / 100.
• Group: df.groupby("major")["score"].mean() answers “average per group”. df.plot(...) hands the table to matplotlib.

Code you should know

import pandas as pd

df = pd.DataFrame({
    "name": ["Ana", "Ben", "Cy", "Dee"],
    "major": ["CS", "Math", "CS", "Bio"],
    "score": [91, 84, 77, 95]
})
print(df.shape, list(df.columns))
print(df.head(2))
print(df["score"].mean(), df["score"].max())
print(df[df["score"] > 80]["name"].tolist())
print(df.sort_values("score", ascending=False).iloc[0]["name"])
print(df["major"].value_counts())

(4, 3) ['name', 'major', 'score']
  name major  score
0  Ana    CS     91
1  Ben  Math     84
86.75 95
['Ana', 'Ben', 'Dee']
Dee
major
CS      2
Math    1
Bio     1

import pandas as pd

df = pd.DataFrame({"city": ["Bozeman", "Bozeman", "Billings", "Billings"], "day": ["Mon", "Tue", "Mon", "Tue"], "high": [61, 68, 70, 74]})
df["high_c"] = (df["high"] - 32) * 5 / 9
print(df.groupby("city")["high"].mean())
print(df.groupby("city")["high_c"].max().round(1))
print(df.loc[df["high"].idxmax(), "city"])

city
Billings    72.0
Bozeman     64.5
Name: high, dtype: float64
city
Billings    23.3
Bozeman     20.0
Name: high_c, dtype: float64
Billings

Worked example

Common mistakes

• df.score > 85 gives a mask of True/False; you still need df[mask] to get rows.
• Confusing loc (labels) with iloc (positions).
• Forgetting that most methods return a new DataFrame; assign the result.