Measure your learning progress weekly with five numbers

Why hours and video completion mislead

A course dashboard showing "72% complete" measures how much you've consumed. You can reach 100% and still be unable to write a join without looking it up. Effort numbers only rise. Output numbers can fall, and that's why they're useful: a drop tells you something needs attention.

The five numbers

1. Build hours. Time spent writing code, formulas or DAX with no tutorial or solution open. Documentation is allowed. Set a floor that fits your life, for example 3 hours a week alongside a full-time job.

2. Blind-task pass rate. Each week, attempt four or five small tasks you haven't seen before: a question from a practice site, a problem set a classmate wrote, or last week's class exercise done again on new data. Record how many you solved without help. This is the closest thing to what an interview measures.

3. Error fix rate. Keep an error log: each time you hit an error you don't understand, write the message and what you were trying to do. At the end of the week, count how many you resolved and understood. A rising pile of unresolved errors is often the first sign you've fallen behind.

4. Deliverables shipped. Finished things another person could open, like a cleaned dataset with notes, a dashboard, or a notebook with a written conclusion. Half-finished work counts as zero. It's harsh, and that's why it works.

5. Explain score. Pick one concept from the week, such as a LEFT JOIN, a confidence interval or a DAX measure, and explain it aloud for 90 seconds as if to a colleague. Score yourself 0 (couldn't do it), 1 (did it, but vaguely) or 2 (clear, with an example).

A tracker that flags problems

Keep the log as a CSV and let a short script do the checking, so you don't have to judge your own week. The data here is synthetic.

import csv
import io

# Synthetic six-week log. Replace with your own rows.
LOG = """week,build_hours,blind_tasks_passed,blind_tasks_tried,errors_logged,errors_resolved,deliverables_shipped,explain_score
1,3,1,4,6,4,0,1
2,4,2,4,5,4,1,1
3,2,1,4,7,3,0,1
4,5,3,5,4,4,1,2
5,5,3,5,3,3,0,2
6,6,4,5,4,4,1,2
"""

rows = list(csv.DictReader(io.StringIO(LOG)))
for r in rows:
    for k in r:
        r[k] = int(r[k])
    r["blind_rate"] = r["blind_tasks_passed"] / r["blind_tasks_tried"]
    r["fix_rate"] = r["errors_resolved"] / r["errors_logged"]

def flags(prev, cur):
    out = []
    if cur["build_hours"] < 3:
        out.append("build hours below floor")
    if cur["blind_rate"] < prev["blind_rate"]:
        out.append("blind pass rate fell")
    if cur["fix_rate"] < 0.6:
        out.append("errors piling up unresolved")
    return out

for prev, cur in zip(rows, rows[1:]):
    print(cur["week"], f"blind {cur['blind_rate']:.0%}", f"fix {cur['fix_rate']:.0%}",
          flags(prev, cur) or "ok")

shipped = sum(r["deliverables_shipped"] for r in rows)
print("deliverables shipped in six weeks:", shipped)
assert shipped == 3
assert flags(rows[1], rows[2]) == ["build hours below floor", "blind pass rate fell",
                                   "errors piling up unresolved"]

In the sample, week 3 raises all three flags at once: build hours fell to 2, the blind pass rate dropped from 50% to 25%, and fewer than half the errors were resolved. That pattern, where all three slip together, usually has one cause outside the course, such as a busy week at work, an illness or a family event. Look for it before blaming your ability. Weeks 4 to 6 recover, and the blind pass rate reaches 80%.

Reading the numbers together

Keep the tasks honest

The blind-task rate is only as good as the tasks. Three rules keep it meaningful:

• Never retry the same task in the same month. If you remember the answer, you're testing memory.
• Mix difficulty: roughly one task you expect to pass easily, two at your level and one that stretches you.
• Time-box each task, for example to 20 minutes. Solving it in 70 minutes after searching around is a partial pass. Count it as half.

Share one line with someone

Once a week, send your five numbers to a classmate, mentor or friend: "Week 6: 6 build hours, 4/5 blind, 4/4 errors fixed, 1 shipped, explain 2." You don't need a reply. Knowing someone will see the numbers is usually enough to keep you honest.

When your blind pass rate stays high for a few weeks, take the fuller readiness check in am I ready to apply for data analyst jobs. The Data Analyst programme sets weekly deliverables that fit straight into number four.