Data Analyst Portfolio Guide: What Recruiters Want to See

# <Name> - Data Analyst Portfolio

SQL | Excel | Power BI | Python

## Featured projects
1. **Repeat Purchase Drivers in Retail** - found repeat orders fell
   among customers with late first deliveries (SQL, Power BI)
2. **Clinic No-show Analysis** - identified time slots with the most
   missed appointments (Python, Excel)
3. **Delivery KPI Dashboard** - daily on-time view by city (Power BI)

Contact: email | LinkedIn

import re

titles = ["Project1", "EDA_final_v2", "Repeat Purchase Drivers in Retail",
          "sales analysis", "Clinic No-show Analysis"]
for t in titles:
    weak = bool(re.search(r"(?i)project\d|final|eda_|^\w+ analysis$", t))
    print(f"{t:36} {'rename: say what you found or studied' if weak else 'ok'}")

-- Which appointment slots have the highest no-show rate?
-- Grain check: appointments has one row per appointment.
WITH slot_stats AS (
  SELECT slot,
         COUNT(*)                                        AS appointments,
         SUM(CASE WHEN status = 'NO_SHOW' THEN 1 ELSE 0 END) AS no_shows
  FROM appointments
  WHERE appt_date BETWEEN '2026-04-01' AND '2026-09-30'
  GROUP BY slot
)
SELECT slot, appointments, no_shows,
       ROUND(100.0 * no_shows / appointments, 1) AS no_show_pct
FROM slot_stats
WHERE appointments >= 50          -- ignore tiny groups
ORDER BY no_show_pct DESC;

import pandas as pd

appts = pd.DataFrame({
    "appt_id": [1, 2, 3, 3, 4],
    "slot": ["8-10", "8-10", "10-12", "10-12", "12-2"],
    "status": ["NO_SHOW", "DONE", "DONE", "DONE", "NO_SHOW"],
})
print("rows:", len(appts), "unique ids:", appts["appt_id"].nunique())
clean = appts.drop_duplicates("appt_id")
print("after de-duplication:", len(clean))
print(clean.groupby("slot")["status"].apply(lambda s: (s == "NO_SHOW").mean()).round(2))

# pip install requests
import requests

links = {
    "Retail project repo": "https://github.com/<you>/repeat-purchase",
    "Dashboard": "https://app.powerbi.com/view?r=<public-link>",
}
for name, url in links.items():
    if "<" in url:
        print(f"{name}: replace placeholder")
        continue
    try:
        code = requests.get(url, timeout=10).status_code
        print(f"{name}: {'OK' if code < 400 else f'BROKEN {code}'}")
    except requests.RequestException as e:
        print(f"{name}: ERROR {e}")

import re

writeup = """Late deliveries caused customers to stop buying.
Customers with late first orders had lower repeat rates in this dataset."""

causal = re.findall(r"(?i)\b(caused|causes|proves|guarantees)\b", writeup)
if causal:
    print("Causal words found:", causal)
    print("Consider: 'is associated with', 'was linked to', 'suggests'.")
else:
    print("no overclaiming words found")

Score each item 0-2 (0 = missing, 1 = partly, 2 = strong)

Clear landing page with featured projects     ___
Project titles describe the question          ___
One-line summary with a finding               ___
Business question at the start                ___
Cleaning steps documented                     ___
SQL readable and correct                      ___
Charts answer the question                    ___
Recommendation and limitations                ___
All links work                                ___
Relevant to the target role                   ___
Total (out of 20)                             ___

Ask a friend to spend two minutes on your portfolio, then answer:

1. What kind of analyst role is this person aiming for?
2. What was the most interesting finding?
3. Which tools do they use?
4. Was anything confusing or broken?

If they cannot answer 1-3, simplify the landing page.

For each project, prepare one 'hook' a recruiter may ask about:

Project: Repeat Purchase Drivers
Hook: Why repeat orders fell among late first deliveries
Evidence: Cohort table + delivery time chart
Caveat: Correlation, not proof of cause
Next step: Suggest a delivery-speed test for first orders