01Dashboarding
Messy data to an executive KPI dashboard
Took messy datasets through cleaning and feature engineering and turned them into clear KPIs and visuals for an executive audience.
Turn on JavaScript to explore this dashboard.
Open to new roles & collaborations
Data Analyst & Insights Specialist
I turn messy data into clear, actionable insights: cleaning and analyzing it with SQL and Python, then building Power BI and Excel dashboards that drive business decisions.
Kenya · gracenganga0742@gmail.com · +254 769 589 467 · linkedin.com/in/grace-nganga-359063379 · gracecv.vercel.app
I'm a data analyst in Kenya. I take the messy part of the job (duplicate rows, blank cells, one region spelled three ways) and turn it into numbers people can trust, and act on.
I work end to end: collecting and cleaning data with SQL and Python, agreeing on the metrics with the people who will use them, then building Power BI and Excel dashboards that tell the story plainly.
The part I enjoy most is the partnership: working with stakeholders to define the right metrics, then validating every result so the recommendation is one they can trust and actually use.
Right now I'm exploring how AI can elevate analytics: using LLMs to speed up data exploration, automate documentation, and prototype predictive workflows that plug into my SQL/Python stack.
Away from the data, you'll find me working through logic puzzles or out cycling.
02Work
Four projects, each with a live demo built on sample data, so you can try the kind of deliverable I build: filter it, forecast with it, follow the data through the pipeline, prioritize with it.
01Dashboarding
Took messy datasets through cleaning and feature engineering and turned them into clear KPIs and visuals for an executive audience.
Turn on JavaScript to explore this dashboard.
02Forecasting
A forecasting workflow that blends classic analytics with AI assistance, with every result validated before it is published.
Turn on JavaScript to explore this forecast.
03Data pipeline
Ingestion, transformation and reporting in one flow, focused on clear metric definitions and decision-ready insights for business users.
Ingest. Three sources land in one staging table with the same columns, so everything downstream reads from one place.
-- Stack the three sources into one staging table
INSERT INTO staging.orders
SELECT order_id, order_date, region, amount, 'pos' AS source, loaded_at
FROM raw.pos_orders
UNION ALL
SELECT order_id, order_date, region, amount, 'web', loaded_at
FROM raw.web_orders
UNION ALL
SELECT invoice_no, txn_date, branch, total, 'quickbooks', loaded_at
FROM raw.qb_invoices;
Clean. Keep the latest copy of every order, standardize dates and region names, and flag missing amounts instead of guessing them.
-- Keep the latest copy of each order, then standardize
WITH ranked AS (
SELECT *,
ROW_NUMBER() OVER (PARTITION BY order_id
ORDER BY loaded_at DESC) AS rn
FROM staging.orders
)
SELECT order_id,
CAST(order_date AS DATE) AS order_date,
INITCAP(TRIM(region)) AS region,
NULLIF(amount, 0) AS amount,
amount IS NULL AS needs_review
FROM ranked
WHERE rn = 1;
Transform. Roll orders up to a monthly table per region and add the features the dashboard and forecast need.
import pandas as pd
orders = pd.read_sql("SELECT * FROM clean.orders", conn)
orders["month"] = orders["order_date"].dt.to_period("M")
orders["is_repeat"] = orders.duplicated("customer_id")
monthly = (
orders.groupby(["month", "region"])
.agg(revenue=("amount", "sum"),
orders=("order_id", "nunique"),
repeat_rate=("is_repeat", "mean"))
.reset_index()
)
monthly["aov"] = monthly["revenue"] / monthly["orders"]
Report. Every KPI is one documented measure, so "revenue" means the same thing on every page and in every meeting.
Revenue YoY % =
VAR CurrentRevenue = [Revenue]
VAR PriorRevenue =
CALCULATE ( [Revenue], SAMEPERIODLASTYEAR ( 'Date'[Date] ) )
RETURN
DIVIDE ( CurrentRevenue - PriorRevenue, PriorRevenue )
04Productivity analytics
A small tool that tests two ideas from Brian Tracy's Eat That Frog! on a real to-do list: the 80/20 rule and the ABCDE method. It shows which few tasks carry most of the value, and which one to do first.
Method from Eat That Frog! by Brian Tracy (opens in a new tab), Berrett-Koehler.
Turn on JavaScript to use the prioritizer.
03Process
Step through a tiny sample dataset the same way I work through real ones. Every number below is computed live from these nine rows.
Turn on JavaScript to step through the sample data.
04Skills
Efficient queries for cleaning, joining and aggregating data, building reliable datasets that power BI reports and analytics models.
Data wrangling, feature engineering and analysis, plus reusable scripts for automation and exploratory work.
Clear dashboards and charts that surface trends and KPIs, designed for readability, context and the people who use them.
Connect and reconcile financial data, map charts of accounts, and export clean datasets that turn bookkeeping into decision-ready reports.
Using LLMs to speed up data exploration, automate documentation and prototype predictive workflows on a SQL/Python stack.
Partnering with stakeholders to define the right KPIs, validate results and turn findings into practical recommendations.
05Contact
I'm always open to new challenges and collaborations. Reach me by email, phone or WhatsApp, or connect on LinkedIn.
gracenganga0742@gmail.com