A Sales Dashboard
September 14, 2019
The purpose of this post is to present a typical sales analysis that might serve as a starting point for the task of analyzing a firm’s sales data. A sample dataset from Kaggle and the latest versions of Solr and Banana(1) will be used for that purpose.
As often required, the dataset needs a bit of pre-processing, such as feature transformation or column name changes, before it can be indexed.
First, an ISO-8601 date field needs to be added. Since the dataset size is small, a spreadsheet processor can be used to insert that column using a simple formula like: LEFT(F2;10) & "T00:00:00Z".
In order to take advantage of the country field when analyzing the dataset on a map, it needs to be converted to standard 2-letter code.(2) Grab a country code mapping and insert it as a new spreadsheet in the same workbook. Then, using a formula like VLOOKUP(V2;'Sheet2:A:B';2), corresponding country codes can be obtained.(3)
The third and last step is to update column names to work with dynamic fields feature of Solr as follows:(4)
ordernumber_i
quantityordered_i
priceeach_f
orderlinenumber_i
sales_f
orderdate_s
orderdate_dt
status_s
qtr_id_i
month_id_i
year_id_i
productline_s
msrp_i
productcode_s
customername_s
phone_s
addressline1_s
addressline2_s
city_s
state_s
postalcode_s
country_s
country_code_s
territory_s
contactlastname_s
contactfirstname_s
dealsize_s
Now, the dataset can be indexed into Solr using the post command in a collection named sales. If you have Docker installed, docker may be used to quickly run Solr and Banana, check this post for more details. Open Banana and load the pre-configured dashboard by clicking Load on the top right and pasting the dashboard Gist URL.
Most of the panels are self-explanatory and interactive: clicking an element reflects on the whole dashboard. On the fifth row however, the quantity distribution panel displays quantity frequencies. It’s noted that quantity is almost equally distributed between 20 and 50 per order. Many panels can be added to get more insights though.