pivot-table

A basic pivot table representing a store sales grouped in categories and subcategories in different states.

Download

Download Zip folder from github or clone from github using the command: git clone https://github.com/GeekSilver/pivot-table.git

Install Dependencies

Run the command npm install or yarn install in root directory.

Build

Run the command npm run build or yarn build This build the app for production to the build folder.

Run the App

Ensure you have node installed in your PC. Run the command npm install -g serve to install serve. Then run serve -s build to serve the static site on port 5000. You can adjust the port where the build app runs to a different port using -l --listen flags. Run the command. serve -s build -l <different-port> You can as well run serve -h to get full list of available options.

Architectural Overview

• Normalize data into Entities (arrays of a single entity type) with the primary keys being the array indices.
• Build relationships between different Entities through Foreign Keys (have a primary key of one entity become a column dimension in another entity).
• Create Composite Keys (primary keys made up of more than one dimension) for easier filtering during summation or aggregation functions.
• Include an hook option to load data fetched from API's. The hook has a url parameter. You can assign the JSONData prop in App.js the value returned by the hook.

Assumptions

• Converted a section of the provided JSON data to a JS Array constant which made it easier for development as the provied JSON was 'too large'.
• Rounded off summation results to nearest whole number.
• This made the precision of sum to be close to but not equal to the precise result which in most cases was a floating point figure.
• Assumed category and sub-category dimensions to be constant row dimensions (of prime interest).
• Metric can be easily configured by code. E.g order.sales can be swapped for order.profit in the entire project files to tabulate profit.

Next Steps

If this were a real project I would:

• Ensure better precision using consistent floating point degree of accuracy e.g using toFixed(2).
• Add a more convenient way of configuring metric.
• Loosen the tightly coupled category and sub-category row implementation to allow for dimension configuring by code using a dimension.config.js file.
• Use more property-agnostic way of calculating sum i.e drop if ... else statements tied to few properties in favour of object destructuring to obtain properties of interest dynamicaly.
• Polish up the User Inteface with attention to detail including better support for accessibility and refactor code to extract better components.
• Accommodate new requirements if any.