Why Every Business Analyst Should Learn pandas in 2026

import pandas as pd

orders = pd.DataFrame({
    "order_id": [1, 2, 3, 4, 5, 6],
    "region": ["North", "North", "South", "South", "East", "East"],
    "promised": pd.to_datetime(["2026-11-01"] * 6),
    "delivered": pd.to_datetime(["2026-11-01", "2026-11-03", "2026-11-01",
                                 "2026-11-01", "2026-11-02", "2026-11-01"]),
})
orders["late"] = orders["delivered"] > orders["promised"]
print("overall late share:", round(orders["late"].mean() * 100), "%")
print(orders.groupby("region")["late"].mean().mul(100).round())

import pandas as pd

df = pd.read_excel("complaints_export.xlsx")    # pip install openpyxl
print("rows:", len(df))
print("duplicate complaint IDs:", df["complaint_id"].duplicated().sum())
print("missing values per column:")
print(df.isna().sum()[df.isna().sum() > 0])
print("date range:", df["created_at"].min(), "to", df["created_at"].max())

import pandas as pd

events = pd.DataFrame({
    "case": ["A", "A", "A", "B", "B", "B"],
    "step": ["received", "verified", "approved"] * 2,
    "ts": pd.to_datetime(["2026-11-01 09:00", "2026-11-02 15:00", "2026-11-03 10:00",
                          "2026-11-01 11:00", "2026-11-01 16:00", "2026-11-04 12:00"]),
})
wide = events.pivot(index="case", columns="step", values="ts")
wide["verify_hours"] = (wide["verified"] - wide["received"]).dt.total_seconds() / 3600
wide["approve_hours"] = (wide["approved"] - wide["verified"]).dt.total_seconds() / 3600
print(wide[["verify_hours", "approve_hours"]])
print("median hours per step:")
print(wide[["verify_hours", "approve_hours"]].median())

import pandas as pd

cases = pd.DataFrame({
    "channel": ["online", "online", "branch", "branch", "online", "branch"],
    "total_hours": [20, 28, 60, 75, 24, 58],
})
summary = (cases.groupby("channel")["total_hours"]
                .agg(["count", "median", "max"]))
print(summary)
# Use the median to describe typical cases; mention the max as a risk.

import pandas as pd

tickets = pd.DataFrame({
    "issue": ["OTP not received", "Upload failed", "OTP not received",
              "Wrong address", "Upload failed", "OTP not received"],
    "minutes_to_resolve": [12, 30, 15, 8, 25, 10],
})
ranked = (tickets.groupby("issue")
                 .agg(tickets=("issue", "size"), total_minutes=("minutes_to_resolve", "sum"))
                 .sort_values("total_minutes", ascending=False))
print(ranked)
# Use this table to justify which requirement to prioritise first.

import sys
import pandas as pd

def weekly_report(path: str) -> None:
    df = pd.read_csv(path, parse_dates=["created_at"])
    last_week = df[df["created_at"] >= df["created_at"].max() - pd.Timedelta(days=7)]
    summary = last_week.groupby("status").size().rename("cases")
    summary.to_csv("weekly_summary.csv")
    print(summary)

if __name__ == "__main__":
    weekly_report(sys.argv[1] if len(sys.argv) > 1 else "cases.csv")
# Usage: python weekly_report.py cases.csv

read_csv / read_excel     load exports
head, shape, dtypes       first look
isna().sum()              missing values
duplicated()              duplicate rows
to_datetime               fix dates
str.strip / str.title     clean text
filter with df[cond]      select rows
groupby + agg             summaries
value_counts              frequencies
pivot / pivot_table       reshape
merge                     join two exports
sort_values               rank
date differences          durations
to_csv / to_excel         share results

Week 1  Install Python (or use a notebook), load Excel and CSV,
        check missing values and duplicates
Week 2  Clean text and dates; filter and sort
Week 3  groupby, pivot tables, merge two exports
Week 4  Build one evidence notebook for a real process question
        and a weekly report script

Question: ______________________________
Data: export name, date range, rows
Checks: duplicates, missing values, date range confirmed
Finding: ______________________________
Evidence: table or chart
Impact on requirements: which story or priority this supports
Caveats: what the data does not show