Data Scientist Needed for Tax Audit Risk Prediction Model (Logistic Regression & Risk Analysis)
Budget: $750 – $1,500 AUD
Project Overview:
We are looking for an experienced data scientist to develop a logistic regression-based audit risk prediction model using internal company data and external industry data. The goal is to predict the likelihood of a tax audit occurring based on various industry, company size, claim-related factors and other factors.
This model will be used across three main R&D periods, which vary in data size and available features, but share the same objective—**assessing tax audit risk.**
Instead of building three separate models, we aim to develop one core model that adapts to each R&D specific characteristics.
-? Scope of Work**
1️⃣ Organizing and Preparing the Provided Data
- We will provide the structured data for the 3 R&D periods.
- You will need to review, clean, and format the data so it’s ready for analysis.
- Handle any inconsistencies or missing values in a logical way (e.g., filling gaps, adjusting formats)
2️⃣ Identifying the Key Factors That Influence Tax Audit Risk
- Analyze for each of the R&D periods, the data to understand which factors affect the likelihood of an audit.
- Example:
- Does company size impact audit risk?
- Are certain industries audited more often?
3️⃣ Building a Model That Works Across Different R&D Periods
- Since we have three different R&D periods, the model should recognise the differences in data while still using the same logic to predict audit risk.
- A special **"R&D Period" label** should be added so the model knows which dataset it is analyzing.
- The model should be flexible enough to adjust its predictions based on each period’s data.
4️⃣ Training & Testing the Model to Ensure It’s Reliable
- Develop a logistic regression model (or suggest an alternative if it works better).
- Train the model using the historical tax audit data provided by us.
- Evaluate the model’s accuracy and reliability by testing it on different R&D periods.
5️⃣ Delivering the Final Model & Insights
- Provide a fully trained and tested model that we can use to assess tax audit risk.
- Deliver a clear report per R &D period with the following info:
- The most important risk factors.
- How the model makes its predictions.
- Any patterns or trends found in the data.
- Ensure the model is well-documented so it can be used, updated, or expanded in the future.
What We Will Provide
✅ Structured data for each R&D period
✅ Descriptions of key variables and data points to guide analysis.
✅ Business context on how the model will be used. (I.e. we will share the key hypothesis we want to test for each R&D period)
What We Expect From the Freelancer:
✅ A trained predictive model for tax audit risk.
✅ A report explaining key findings and recommendations.
✅ Documentation on how to use and maintain the model.
Required Skills
✅ Machine Learning & Predictive Modeling
- Experience with logistic regression (binary, multinomial, or ordinal).
- Ability to evaluate model performance using accuracy, AUC-ROC, precision-recall, etc.
✅ Data Analysis & Feature Engineering
- Strong understanding of financial, insurance, or tax risk modeling (big plus ).
- Experience with feature selection to identify key predictors of audit risk.
- Ability to handle missing or incomplete data intelligently.
✅ Programming & Tools
- Proficiency in Python (Pandas, NumPy, Scikit-learn, Statsmodels, Matplotlib/Seaborn).
- SQL knowledge (nice to have)
- Experience with Jupyter Notebooks for reporting and documentation.
✅ Communication & Documentation
- Ability to clearly explain findings and insights in a non-technical way.
- Experience delivering well-documented models so they can be used and updated easily.
Timeline & Budget
⏳ Estimated Turnaround: 1 week - please include an estimated timeline in you proposal
Budget: Open to discussion – please provide your bid
How to Apply
Please include:
✔️ A brief outline of your approach.
✔️ Estimated time for completion.
✔️ Relevant experience or past projects.
Looking forward to working with an expert to unlock valuable insights from our data! ?
We are looking for an experienced data scientist to develop a logistic regression-based audit risk prediction model using internal company data and external industry data. The goal is to predict the likelihood of a tax audit occurring based on various industry, company size, claim-related factors and other factors.
This model will be used across three main R&D periods, which vary in data size and available features, but share the same objective—**assessing tax audit risk.**
Instead of building three separate models, we aim to develop one core model that adapts to each R&D specific characteristics.
-? Scope of Work**
1️⃣ Organizing and Preparing the Provided Data
- We will provide the structured data for the 3 R&D periods.
- You will need to review, clean, and format the data so it’s ready for analysis.
- Handle any inconsistencies or missing values in a logical way (e.g., filling gaps, adjusting formats)
2️⃣ Identifying the Key Factors That Influence Tax Audit Risk
- Analyze for each of the R&D periods, the data to understand which factors affect the likelihood of an audit.
- Example:
- Does company size impact audit risk?
- Are certain industries audited more often?
3️⃣ Building a Model That Works Across Different R&D Periods
- Since we have three different R&D periods, the model should recognise the differences in data while still using the same logic to predict audit risk.
- A special **"R&D Period" label** should be added so the model knows which dataset it is analyzing.
- The model should be flexible enough to adjust its predictions based on each period’s data.
4️⃣ Training & Testing the Model to Ensure It’s Reliable
- Develop a logistic regression model (or suggest an alternative if it works better).
- Train the model using the historical tax audit data provided by us.
- Evaluate the model’s accuracy and reliability by testing it on different R&D periods.
5️⃣ Delivering the Final Model & Insights
- Provide a fully trained and tested model that we can use to assess tax audit risk.
- Deliver a clear report per R &D period with the following info:
- The most important risk factors.
- How the model makes its predictions.
- Any patterns or trends found in the data.
- Ensure the model is well-documented so it can be used, updated, or expanded in the future.
What We Will Provide
✅ Structured data for each R&D period
✅ Descriptions of key variables and data points to guide analysis.
✅ Business context on how the model will be used. (I.e. we will share the key hypothesis we want to test for each R&D period)
What We Expect From the Freelancer:
✅ A trained predictive model for tax audit risk.
✅ A report explaining key findings and recommendations.
✅ Documentation on how to use and maintain the model.
Required Skills
✅ Machine Learning & Predictive Modeling
- Experience with logistic regression (binary, multinomial, or ordinal).
- Ability to evaluate model performance using accuracy, AUC-ROC, precision-recall, etc.
✅ Data Analysis & Feature Engineering
- Strong understanding of financial, insurance, or tax risk modeling (big plus ).
- Experience with feature selection to identify key predictors of audit risk.
- Ability to handle missing or incomplete data intelligently.
✅ Programming & Tools
- Proficiency in Python (Pandas, NumPy, Scikit-learn, Statsmodels, Matplotlib/Seaborn).
- SQL knowledge (nice to have)
- Experience with Jupyter Notebooks for reporting and documentation.
✅ Communication & Documentation
- Ability to clearly explain findings and insights in a non-technical way.
- Experience delivering well-documented models so they can be used and updated easily.
Timeline & Budget
⏳ Estimated Turnaround: 1 week - please include an estimated timeline in you proposal
Budget: Open to discussion – please provide your bid
How to Apply
Please include:
✔️ A brief outline of your approach.
✔️ Estimated time for completion.
✔️ Relevant experience or past projects.
Looking forward to working with an expert to unlock valuable insights from our data! ?