How Pandas Helps You Land a Software Tester Job Faster

Software Testing · Data Validation

How Pandas Helps You Land a Software Tester Job Faster

SECTION 01Validating data against business rules

Business analysts and product owners usually define these rules in requirement documents, and testers turn them into checks. Every data field has rules: required, unique, within a range, in a format. Pandas lets you turn each rule into a one-line check across all rows instead of sampling a few by eye.

import pandas as pd

df = pd.read_csv("customers_export.csv", dtype={"pin": str})
checks = {
    "missing_customer_id": df["customer_id"].isna(),
    "duplicate_customer_id": df["customer_id"].duplicated(keep=False),
    "bad_pin": ~df["pin"].fillna("").str.fullmatch(r"\d{6}"),
    "bad_discount": ~df["discount"].between(0, 50),
}
for name, mask in checks.items():
    print(f"{name:25} {int(mask.sum())} rows")

failed = []
for name, mask in checks.items():
    rows = df[mask].copy()
    rows["failed_rule"] = name
    failed.append(rows)
pd.concat(failed).to_csv("validation_defects.csv", index=False)
print("Defect evidence saved to validation_defects.csv")

SECTION 02Reconciling source and target data

Migrations and data pipelines are high-risk changes because errors can be silent: the app still works, but some records are wrong or missing. After a migration or ETL job, testers must prove nothing was lost or changed by mistake. Pandas compares two datasets row by row and value by value, and gives you evidence you can attach to the test report.

import pandas as pd

src = pd.read_csv("legacy_orders.csv")
tgt = pd.read_csv("new_system_orders.csv")
print("rows:", len(src), "->", len(tgt))

m = src.merge(tgt, on="order_id", how="outer",
              suffixes=("_src", "_tgt"), indicator=True)
print(m["_merge"].value_counts())
m[m["_merge"] != "both"].to_csv("missing_records.csv", index=False)

both = m[m["_merge"] == "both"]
for col in ["amount", "status", "customer_id"]:
    diff = both[both[f"{col}_src"] != both[f"{col}_tgt"]]
    print(f"{col:12} {len(diff)} mismatched rows")
print("total src:", round(src["amount"].sum(), 2),
      "total tgt:", round(tgt["amount"].sum(), 2))

SECTION 03Verifying reports and KPIs

Managers make decisions from these numbers, so a small calculation error can have a big business impact. Dashboards and reports are often tested only visually. Recomputing their key numbers from raw data with Pandas catches calculation and filter bugs that a screenshot comparison would never notice.

import pandas as pd

orders = pd.read_csv("orders.csv", parse_dates=["created_at"])
orders["month"] = (orders["created_at"]
                   .dt.tz_localize("UTC")
                   .dt.tz_convert("Asia/Kolkata")
                   .dt.to_period("M").astype(str))
expected = orders.groupby("month")["amount"].sum().round(2)
print(expected)

report = pd.read_excel("monthly_revenue_report.xlsx", index_col="month")["revenue"]
compare = pd.DataFrame({"expected": expected, "report": report})
compare["diff"] = (compare["expected"] - compare["report"]).round(2)
print(compare[compare["diff"].abs() > 0.01])

SECTION 04Reusable data test templates

Wrap your checks in pytest so they run like any other test suite and produce clear pass or fail results. Once they are tests, they can run automatically in CI every time a new data extract or build is ready, which turns one-off checks into lasting protection. For a related walkthrough, see Pandas Skills for QA Tester Jobs.

# tests/test_orders_data.py
def test_no_missing_order_ids(orders):
    assert orders["order_id"].notna().all()

def test_order_ids_unique(orders):
    assert not orders["order_id"].duplicated().any()

def test_amounts_not_negative(orders):
    assert (orders["amount"] >= 0).all()

def test_status_values_valid(orders):
    allowed = {"placed", "shipped", "delivered", "cancelled"}
    assert set(orders["status"].unique()) <= allowed

# conftest.py
import pandas as pd
import pytest

@pytest.fixture(scope="session")
def orders():
    return pd.read_csv("data/orders.csv", dtype={"order_id": str})

Data test summary (attach to test report)
Dataset: orders, 48,210 rows, build 2.14
Rules checked: 12   Passed: 11   Failed: 1
Failure: 37 orders with negative amounts (evidence attached)
Reconciliation: 48,210 -> 48,210 rows, totals match
Recommendation: fix negative amounts before sign-off

SECTION 05Data testing habits

These habits keep data checks reliable and your findings credible with developers and data teams, who will often double-check your numbers. A data defect that turns out to be a testing mistake damages trust quickly, so care in the setup pays off every time. You may also find Tableau Skills for QA Tester Jobs and Why QA Testers Should Learn Git and GitHub useful.

• Agree rules in writing — confirm field rules with the business analyst before testing.
• Keep evidence files — save failing rows to CSV and attach them to defects.
• Read keys as strings — protect leading zeros and long numeric IDs.
• Use tolerances for money — round consistently before comparing amounts.
• Never test on live customer data casually — use approved test extracts and mask personal fields.

SECTION 06Interview Q&A — pandas for software testers

SECTION 07Test yourself — pandas for software testers quiz

SECTION 08Frequently asked questions

SECTION 09Related reads