AdTech Data Analysis & Experimentation Task (SQL + Python + Data Visualization) -- 2
Budget: ₹999 – ₹3,000 INR
I’m looking for an experienced Data Analyst / Data Scientist with AdTech or performance analytics background to support a short-term data analysis task.
The project involves analysing a provided impression-level dataset (CSV) containing user, device, and contextual information over a one-month period. The goal is to extract insights, assess monetisation performance, and define features for a conversion prediction use case.
This is a one-off analytical task, expected to take 3–4 hours for someone experienced.
Scope of Work
The work includes:
SQL-based analysis
Calculate IPM (Installs per Mille) per country using SQL only.
Identify top-performing segments.
Performance & risk analysis
Assess risks of scaling budget on top-performing demand apps.
Highlight biases, stability concerns, and data limitations.
Feature definition for conversion prediction
Propose:
3 contextual features
3 behavioural features
Explain why they are useful.
Feature engineering
Write Python code to calculate the features.
Explain risks such as leakage, bias, or instability.
Visualization
Create one presentation-ready chart showing how a contextual and behavioural feature relate to conversion rate.
Audience: Head of Product / non-technical stakeholders.
Delivery
Jupyter Notebook or ZIP containing:
Code
Clear written explanations
Final visual
No raw dataset should be included in the final delivery.
The project involves analysing a provided impression-level dataset (CSV) containing user, device, and contextual information over a one-month period. The goal is to extract insights, assess monetisation performance, and define features for a conversion prediction use case.
This is a one-off analytical task, expected to take 3–4 hours for someone experienced.
Scope of Work
The work includes:
SQL-based analysis
Calculate IPM (Installs per Mille) per country using SQL only.
Identify top-performing segments.
Performance & risk analysis
Assess risks of scaling budget on top-performing demand apps.
Highlight biases, stability concerns, and data limitations.
Feature definition for conversion prediction
Propose:
3 contextual features
3 behavioural features
Explain why they are useful.
Feature engineering
Write Python code to calculate the features.
Explain risks such as leakage, bias, or instability.
Visualization
Create one presentation-ready chart showing how a contextual and behavioural feature relate to conversion rate.
Audience: Head of Product / non-technical stakeholders.
Delivery
Jupyter Notebook or ZIP containing:
Code
Clear written explanations
Final visual
No raw dataset should be included in the final delivery.