Browse categories
Explore
Fiverr Pro
English
$
USD
I will do statistical analysis and hypothesis testing on your data in python
Data Analyst
I turn messy spreadsheets into clear reports and dashboards that show you what to do next.
Mechatronics engineer, university lecturer and former Data & Reporting Analyst, I have spent years cleaning,...
About this Gig
"Does X actually affect Y, or does it just look that way?"
That is the question this gig answers, with a test that fits your data and an answer you can explain to someone who has never heard of a p-value.
HOW I WORK
First I look at your data and your question and tell you which test is appropriate and why. Assumptions get checked, not assumed. If your data cannot support the question, I say so before you order that answer is free.
WHAT YOU GET
- The right test for your data, with the reason it was chosen
- Assumption checks, effect size and confidence, not just a p-value
- A chart that shows the result
- A plain-English verdict: what the data supports, what it does not, and how confident you can be
- The limitations stated openly, including where correlation is not causation
TYPICAL QUESTIONS
Did the new pricing change average order value? Do customers from channel A behave differently from channel B? Did the A/B test actually win? Is this difference real or is it noise?
TOOLS
Python: pandas, numpy, scipy, matplotlib.
NOT THIS GIG
Academic assignments, theses and coursework. Predictive models and machine learning. Message me first if you are unsure.
Analysis type:
Statistical analysis
•
Descriptive analysis
Expertise:
Business Insights
•
Statistics
Programming language:
Python
•
SQL
My Portfolio
FAQ
Which test will you use?
Whichever one your data justifies. That depends on distribution, sample size and what you are comparing. I tell you the choice and the reason before I run anything — I do not pick a test because it gives a nicer result.
What if the result is not significant?
You get that answer, clearly stated, and it is worth exactly as much as a positive one. "The difference you are seeing is probably noise" saves more money than most positive findings make.
How much data do I need?
Send what you have and I will tell you before you order. Small samples are not automatically useless, but they change which test is valid and how confident the answer can be.
Can you prove that X caused Y?
Almost never, and anyone who promises that is selling you something. Observational data shows association. I will tell you clearly what would be needed to claim causation.
Do you do machine learning or forecasting?
Not in this gig. That is a different kind of project with different risks, and I would rather say no than deliver a model I cannot stand behind.
