davide-rafaschieri / repository

powerbi_adventure_works

AdventureWorks is a fictitious multinational manufacturing company. The data set used for this report is from the AdventureWorks2014 version. The goal of this project is to demonstrate how building powerful and beautiful (I hope :)) reports in Power BI by following the data visualization best practices and the data modelling patterns, showing the awesome features of Power BI like Report Page Tooltip, Calculated Groups, Forecasting, What-If parameters, Time Intelligence Functions, complex DAX code, custom charts and how it is easy to integrate Power BI with Python/R script in order to build machine learning models and so creating advanced and predictive analytics reports. Reports Pages are as follow: - P&L Overview: Profit & Loss Report, in which I analyzed the economic status of the company from the point of view of Revenues and Expenditures. In this report page I compared the trend of actual amount and budget amount, unfortunately for 2011 year only (missing data for other years). You can notice the powerful feature of “report page tooltip” in this page, by hovering over the bar charts. - Internet Sales Overview: in this report page I analyzed the sales of products to customers via Internet. You can notice the forecasting feature of Power BI and the use of what-if parameters in order to build advanced analytics reports. Also I made use of calculated groups (for further info about visit https://www.sqlbi.com/articles/introducing-calculation-groups/). - Reseller Sales Overview: similar to the previous one, but this time the focus is on the sales of products to resellers. Things to notice in this report page are the comparison of current sales vs last year sales, the comparison of actual sales amount vs budget/quota sales amount, and the running total and moving/running average charts. - Product Inventory Overview: inventory management analysis of the company. I analyzed the current stock on hold vs the recent sales revenue, showing the value of current stock, the current units in stock and the stock ratio. - Products Overview: in this report page I analyzed the products by category, color, size, “ABC” class too. Two very interesting and powerful things to notice: the basket analysis using DAX and the basket/association rules analysis using R script and apriori machine learning model. I made use of the custom chart called “Network Chart” to represent the association rules between products. Through the basket analysis the user could know which product is likely to be bought with another one. - Customers Overview: customer analysis by age group, location, status, and so on. The customer could be classified as “Bike Buyer” and “Non Bike Buyer”. So I decided to build a machine learning model (in this case I used XGBoost algorithm in Python) in order to predict if a (new) customer would be a bike buyer or not. The model has a prediction accuracy of about 84% which it is not bad. - Employees Overview: employee analysis by sales location, title, age. Thing to notice is the ribbon chart through which I can analyze the performance of employees over time. - Reseller Overview: analysis of the performance of the resellers. Important dimensions to analyze are the business type, the product line, the reseller location.