Predictive Data Analysis Report
Budget: €30 – €250 EUR
The report should consider a real world problem using a non-trivial dataset (for regression). You may find a suitable dataset in course content or on UCI Machine learning repository. In the repository look for regression datasets, which are multivariate, and with mixed or numerical attribute types (https://archive.ics.uci.edu/ml/datasets.php?format=&task=reg&att=mix&area=&numAtt=&numIns=&type=mvar&sort=nameUp&view=table). There is plenty of interesting datasets for modeling with machine learning.
The report should be well structured and contain at least the following chapters:
- Executive Summary
- Introduction – Explain the business problem, plan to solve it, short summary of data & selected outcomes (KPIS).
- Data Description – Source of data (references), Quality & reliability of data. # parameters, #records, health
- Analysis of mutual relationships among variables (attributes). – Interesting insights from pair-wise correlations. (In human/buisiness language)
- Predictive modeling – report of constructed predictive models (quality/performance) to achieve the business goal.
– Insights – Focus on the insights (confusion matrix or what-ifs) - try to explain the relationships captured by the models, give example insights.
- Conclusions & Future Work (in business language).
- References.
The report should contain a least 10 and at most 25 pages.
You may use DataStories software platform to do the modeling iterations.
The report should be well structured and contain at least the following chapters:
- Executive Summary
- Introduction – Explain the business problem, plan to solve it, short summary of data & selected outcomes (KPIS).
- Data Description – Source of data (references), Quality & reliability of data. # parameters, #records, health
- Analysis of mutual relationships among variables (attributes). – Interesting insights from pair-wise correlations. (In human/buisiness language)
- Predictive modeling – report of constructed predictive models (quality/performance) to achieve the business goal.
– Insights – Focus on the insights (confusion matrix or what-ifs) - try to explain the relationships captured by the models, give example insights.
- Conclusions & Future Work (in business language).
- References.
The report should contain a least 10 and at most 25 pages.
You may use DataStories software platform to do the modeling iterations.