FC Barcelona Sierra Leone

Directions

FC Barcelona Sierra Leone 🇸🇱 Data Scientist | AI Educator | Founder, RiseAfrica Foundation for STEM & Innovation. Learn • Build • Share • Inspire.
(2)

Helping Africans learn AI, Data Science & technology to solve real-world problems.

A customer moved cities. Your revenue report quietly moved with her.Ana placed 3 orders in New York ($220), moved to Aus...
10/16/2026

A customer moved cities. Your revenue report quietly moved with her.

Ana placed 3 orders in New York ($220), moved to Austin on July 1, then ordered once more ($80). Overwrite her city (Type 1) and the report says Austin $300, New York $0. Keep a new row with valid_from and valid_to (Type 2) and it says New York $220, Austin $80. Join on the date range too, or you double count to $600.

Comment SCD2 for the notes.

10/16/2026

Get Percentages in Pandas With One Argument

value_counts() tells you how many. value_counts(normalize=True) tells you what share of the total. Add .mul(100).round(1) for clean percentages.

10/16/2026

Build a Data Portfolio for $0 With 4 Free Tools

You don't need a paid course to show recruiters what you can do. Kaggle for data, Google Colab for analysis, Looker Studio for dashboards and GitHub to share it, all free.

Two models. Both 90% accurate. One of them is dangerous.Model A gives honest probabilities, and its one miss said 45%. M...
10/16/2026

Two models. Both 90% accurate. One of them is dangerous.

Model A gives honest probabilities, and its one miss said 45%. Model B says 99% or 1% about everything, and its one miss said 1% when the answer was yes. Accuracy can't tell them apart. Log loss can: 0.31 for A, 0.47 for B. That one confident miss alone costs 4.6, because the penalty is −ln(p).

Comment LOGLOSS for the notes.

10/16/2026

This SQL Trick Finds Hidden Duplicate Customers

"[email protected]" and "[email protected] " are the same customer, but a plain GROUP BY can count them as two. LOWER + TRIM + HAVING finds the hidden duplicates in one query.

10/16/2026

Why Your SQL Total Says NULL (One-Word Fix)

Salary + bonus should never come back NULL. In SQL, a number plus NULL is NULL, so one missing bonus wipes out the whole total. COALESCE(bonus, 0) fixes it in one word.

Your conversion rate says 0%. Your signups are real.In SQL, an integer divided by an integer stays an integer. SELECT 3/...
10/16/2026

Your conversion rate says 0%. Your signups are real.

In SQL, an integer divided by an integer stays an integer. SELECT 3/4 returns 0. In my test, 126 signups / 4,000 visits came back as 0, and even 100*37/2500 came back as 1 instead of 1.48. Write 1.0 * signups / visits (or CAST one side) and you get the real rates: 3.15%, 9.22% and 1.48%.

Comment INTDIV for the notes.

Your load job failed halfway. The scheduler retried it. Now revenue is doubled.A plain INSERT has no memory. In my test,...
10/15/2026

Your load job failed halfway. The scheduler retried it. Now revenue is doubled.

A plain INSERT has no memory. In my test, 3 orders loaded twice became 6 rows and $475.50 instead of $237.75. Put a key on order_id and use INSERT ... ON CONFLICT DO UPDATE (or MERGE). I ran it three times: still 3 rows, still $237.75.

Comment UPSERT for the notes.

Your model says Billing 2.0, Technical 1.0, Sales 0.1. Those aren't probabilities yet.Softmax makes them probabilities: ...
10/15/2026

Your model says Billing 2.0, Technical 1.0, Sales 0.1. Those aren't probabilities yet.

Softmax makes them probabilities: e^2.0 = 7.389, e^1.0 = 2.718, e^0.1 = 1.105. Divide each by the total (11.21) and you get 65.9%, 24.2% and 9.9%, which add up to 100%. Add 100 to every score and nothing changes, because only the differences matter.

Comment SOFTMAX for the notes.

Your query uses the index. It's still slow.An index on user_id finds the matching rows, then jumps back to the table for...
10/15/2026

Your query uses the index. It's still slow.

An index on user_id finds the matching rows, then jumps back to the table for every single one to read status. In my 500,000-row SQLite test, that returned 49,674 rows in about 103 ms, slower than scanning the whole table (about 80 ms). Index (user_id, status) instead: the plan says USING COVERING INDEX, same rows, about 21 ms.

Comment COVERING for the notes.

Address

Somerset
Jersey City, NJ
08873

Website

Alerts

Be the first to know and let us send you an email when FC Barcelona Sierra Leone posts news and promotions. Your email address will not be used for any other purpose, and you can unsubscribe at any time.

Shortcuts

Share

Category