OnlineRetail dataset, Analysis#3.1 - Power BI

OnlineRetail dataset - MS Power BI, additional analysis #1

Dataset reference: the analyzed dataset - mentioned on DataCamp website - was downloaded from the site of Online retail dataset sharing/owner.

Basic analysis by Power BI of Online Retail dataset revealed simple facts.

In another two projects, I analyzed the dataset with Python and SQL with the help of the in-built visualization of the Notebook offered on DataCamp website and also using matplotlib for special purposes.

The complete dataset and the BI solution is zipped. The package sizes 26 MB and requires installed Microsoft Power BI software (ver 2.130+, made wtih ver 2.131.1203.0).

As a simple step Microsoft (MS) Power BI software is able to load data easily from Excel files. The DataCamp project defined simple questions that could be easily answered in a short time.

Here I demonstrate a visualization that cannot be created neither by DataCamp (or any) online Notebook application even if those regularly have an in-built data visualization tool, nor by offline analysis in Python using the matplotlib module.

The advantage of Power BI software is to plot 2 datasets ("columns") on a 2D graph's X and Y axis but to add a 3rd layer (not axis) of some additional information, strongly connected to the previous datasets. In such case, a completely new viewpoint comes to the picture. Using another advantage of Power BI, the plot can be made dynamic so that selecting some value of the 3rd layer subset of data can be highlighted or specifically selected for visualization and consequently a confusingly large dataset can be easily analyzed according to our needs depending on our selection.

The plot(s) below visualize the frequency of quantities of purchased product amounts with the additional information of purchaser's country. The vertical axis (Y) represents the frequency of well defined purchased amounts (1, 2, 3, ... pieces) versus the purchased amounts (pieces) on the horizontal axis, while different countries are presented with different colors (automatically set by Power BI). Note that both axis are in logarithmic representation therefore the 1-100 distance is the same as the distance of 100 - 10 000 units, but for first understanding it is not important.

Frequency of purchased quantities by Countries (default view)

This is a bit crowded therefore it would be great to have some selector with what we can define subsets in what we are interested in. Power BI has a Slicer option in which the selection categories may be loaded by a simple drag&drop method. In this case 30+ countries appear in the slicer, each countries in a separate box.

Frequency of purchased quantities - all countries plot, with an additional slicer.

In deed, the slicer takes quite a large part, but there is no option to rearrange or minimize internal margin in the boxes. Still it is a really good opportunity to select for example United Kingdom (UK) related data: 

Frequency of purchased quantities - UK

UK purchases has a well defined triangle, which seems to have a sharp lower edge. It is due to the fact thet UK customers has at least 100 times more data entries then other countries, so it is evident that the frequency values cannot take low numbers for low purchased quantities.

Let's see the non-UK purchase quantities:

Frequency of purchased quantities - non-UK countries

Note: based on these plots above the DataCamp 3rd question "is non-UK purchased amounts significantly higher than UK purchsed amounts?" cannot be precisely answered, however we may have an impression. For significant difference analyses statistical t-probe should be utilized (not presented here).

Countries of our interest can be selected for specific analyses:

Purchases as income and refunds related losses on maps

Country-related total values of purchases or returns expressed in quantities or money may be plotted using Power BI maps.

Heat map would be more useful to express the differences of countries' contribution to the whole amounts. Keeping mind that UK entries are extremely higher in the dataset compared to non-Uk countries relative numbers should be plotted. Map not presented.

Additional analysis #2 - Clustering with Power BI

OnlineRetail dataset, Analysis#3 - Power BI

Analysis 3rd attempt using MS Power BI on the DataCamp - OnlineRetail dataset

---

Additional Power BI #1 - Frequency of purchased quantities in countries

Additional Power BI #2 - Clustering with Power BI

---

In the first two runs, I analyzed the dataset with Python and SQL with the help of the in-built visualization of the Notebook offered on DataCamp website and also using matplotlib for special purposes.

Dataset reference: The original dataset (xlsx file) was downloaded from the Dataset owner's site.

The complete dataset and the BI solution is zipped. The package sizes 26 MB and requires installed Microsoft Power BI software (ver 2.130+, made wtih ver 2.131.1203.0).

Microsoft (MS) Power BI software is able to load data easily from Excel (or from multiple files or online databases) and recognize data types automatically so I only had to concentrate on visualization of the required information. I created the simplest figures with minimal style adjustments, as making more elaborate or eye-catching visuals would have taken nearly twice as long to complete the project.

Before the questions and answers, here you find some general information on the data and the dataset:

First of all, in general the most important data for Sales&Management is what are the flagship products? Of course they already know, but this is a proof of properly working queries on a well designed database/dataset. See in table, and in graph format:

It is important to know the dataset size, the amount of countries and Users (website user Customers), see the numbers on the right:

or evidently being UK centered it is also benefitial to know the most returned goods in UK or even better to know the users (identifiers) who have returned the most amount of goods, to extract their list and to target them either by some discount or by feedback questionnaire to understand the reason of returns... with an aim to care with them and regain their trust and loyalty toward our company and our products.

In the order of the questions, let's go through the answers.

1) The most returned items

The number of return related entries in the dataset is around 10 620, sent back by 1590 Customers from 30 countries.

The details of the Returns - general dashboard

Not surprisingly the returned goods in highest amounts are related to UK customers, it is a direct consequence of that the most purchases are originating from there, as well (data not shown).

At the bottom Treemap, UK related transactions are excluded to demonstrate the contribution of different countries in the amount of returned goods (the tile graph may be optimized in size).

2) Profits earned in the UK, on different timescales

A simple demand, easy to answer using the in-built [Month] or [MonthNo], [Quarter] part of DateTime variable. I used 'Online Retail'[InvoiceDate][MonthNo] to define a new column in the data Table. 

In case of weeks the week number can be determinde by the formula

WEEKNUM('Online Retail'[InvoiceDate])

in a new column.

The profit is considered as the sum of the purchases and returns related costs/refunds, so a 'simple' sum of Quantity*Unit price for all transactions in the given time period. Remember that returned quantities are defined with negative integers.

3) Is non-UK purchased amount significantly larger than UK quantities?

First thing to mention; Power BI does not have an in-built (statistical) t-test functionality, so at the moment I provide only such simple graphs and related data that is the base of a t-test. Expert data analyser 

Note that the larger mean value does not prove immediately that non-UK customers buy goods in significantly larger amount. However the plots would indicate so for a professional Data Scientist, the absolut proof can be made by a t-test (not shown here).

In addition to the above-mentioned dataset-related supplementary information, a lot of other facts and conclusion can be extracted from the dataset, but this post is about answering the DataCamp defined 3 questions using Power BI.

OnlineRetail dataset, Analysis#2 - SQL

DataCamp - OnlineRetail dataset,  Analysis #2 - SQL with Notebook

Previously I analyzed the below presented dataset with python with the help of the in-built visualization of the Notebook offered on DataCamp website and also using matplotlib for special purposes. I created another analysis with and MS Power BI.

I have elaborated the answers for the so called Online Retail dataset with SQL queries and zipped all required files (data and explanation as csv files, notebook as ipynb file) GitHub. If you have no access to any Notebook that is compatible* with the ipynb file then a pdf version is available here: SQL queries (0.1 MB) or the complete python and SQL solutions without the coding part Python+SQL (12 MB)

My solution includes additional details and goes beyond the required questions.

The Online retail project (original data source) is presented by DataCamp in steps:

1. Problem definition of E-Commerce Data 
   
   
   
2. Data Dictionary - explaining the content, the data types and the meaning of specific values or signs
   
   
   
   
   
3. "Don't know where to start?" session defines Exploration, Analysis, and Visualisation challenges:
1. Explore: Negative order quantities indicate returns. Which products have been returned the most?
2. Visualize: Create a plot visualizing the profits earned from UK customers weekly, monthly.
3. Analyze: Are order sizes from countries outside the United Kingdom significantly larger than orders from inside the United Kingdom?

With SQL visualization, and statistical analysis is not possible, but the data can be extracted for further analysis. DataCamp website in-built Notebook functionality can be of help in case of visualization, but t-probe statistics is still not doable on a pure SQL basis.

Duck DB's SQL is a bit different from other SQL languages so loading the data already required a specific method: 

Searching for the most returned items during the whole period defined by the dataset:

Preparing data for profit  (in UK) visualization at different timescales:

See the Python-based study or the Power BI solution on the same topic.

* Compatible Notebook readers/editors: DataCamp online Notebook (requires registration) Please note that Jupyter Notebook versions of any kind are not compatible automatically with any kind of SQL interpreters. There are ways (search online for solutions), but I recommend reading the pdf version provided above.

DC OR editable (for myself)

OnlineRetail dataset, Analysis#1 - Python

DataCamp - Online Retail dataset,  Analysis #1 - Python supported by Notebook functionality

I have elaborated the answers for the so called Online Retail dataset and zipped all required files (data and explanation as csv files, notebook as ipynb file) GitHub. If you have no access to any Notebook that is compatible* with the ipynb file then a pdf version is available here: Analysis with Python (10 MB)

My solution includes additional details and goes beyond the required questions.

The Online retail dataset (original source) is presented by DataCamp in steps:

1. Problem definition of E-Commerce Data 
   
   
   
2. Data Dictionary - explaining the content, the data types and the meaning of specific values or signs
   
   
   
   
   
3. "Don't know where to start?" session defines Exploration, Analysis, and Visualisation challenges:
1. Explore: Negative order quantities indicate returns. Which products have been returned the most?
2. Visualize: Create a plot visualizing the profits earned from UK customers weekly, monthly.
3. Analyze: Are order sizes from countries outside the United Kingdom significantly larger than orders from inside the United Kingdom?

Here are some extracted figures.

The monthly income (Quantity * UnitPrice) and repayments (Quantity * UnitPrice). The December - January returns are following the October - December increased pucrhases. "End of the year" craziness. :) 

The 2 figures below are specific visualizations of the "How often each product is returned (frequency) and in what amount?" These help to perceive the outliers: the weakest and the best-sold (with lowest return rate) products in our portfolio.

The t-test result of the hypothesis "non-UK costumers purchase goods in significantly higher amount than UK customers".  Further details can be found in the above-mentioned Notebook or pdf files.

See the SQL-based study or the Power BI solution on the same topic.

* Compatible Notebook readers/editors: DataCamp online Notebook (requires registration) or Anaconda Jupyter Notebook (requires installation) or the online version of Jupyter

DC OR editable (for myself)

DataCamp - excercises

https://www.datacamp.com/

DataCamp is a platform that offers (big) data resources for practicing data processing, analysis, and visualization.

It provides a wide range of Resources, including regularly scheduled Webinars, as well as always available Tutorials, White papers and Podcasts.

If you remember a concept but not the exact code or module to use... it's always useful to have handy Cheat sheets, which are provided for various software and languages like Python, SQL, Excel, bash, and others.

There are also job-related guides, the latest industry news for employees, and discussions on employer hiring needs - ideal for employers looking to fill positions like Data Scientist.

Basic registration and services are free, with options for customization, especially for businesses or educational institutions.

See elaborated examples of Python-based or SQL-based study or the Power BI solution on the topic of "Online Retail" dataset and related questions. The first two were analyzed with the help of the in-built Notebook on the DataCamp / DataLab website, which offered Python, SQL interpretation, and basic in-built visualization tools (as table or graph).

Online Datasets - for Practice

Easy to reach online datasets 

There are plenty of websites that provide smaller or larger datasets that can be used for free or for a certain amount of money to practice DataScience related duties such as accessing data, recognition / understanding data content and data types, data cleaning / conversion / data manipulation (ETL) and finally representation of extracted key information, drawing consequences or making predictions, clustering, segmentation, and so on, depending on the predefined requirements or the possible ways of use of datasets.

It is rarely mentioned but double-checking the extracted information is an inevitable step not to mislead yourself or the stakeholders in a real-life project.

Here you find some websites helping to find datasets:

• Datacamp jump site 
• Google Dataset search
• GeeksForGeeks blog

Here is a short list of such websites that give partially or completely free access to datasets:

• Datacamp - this website offers notebook-based data juggling, tutorials, and education in AI & ML domains
• Kaggle - this website offers competitions, but also tutorials and education in AI & ML domains
• Data.gov - USA Federal Government datasets, of course, the non-confidential part
• Earth data - collected by NASA
• Global Health Observatory Data - for those who like health-related issues or facts

Have fun with the data suiting you the best!

sn - AI & ML for Data

I have found two interesting publications about the required skills and tools for Data Analyst, Data Scientist, Data Engineer, and Machine Learning Engineer positions. They collected skill lists for each position scratching multiple job advertisements for months and used AI (ML) to extract data and also for data clustering.

The required skill lists have nothing surprising, but I found elegant the way they did it.

One important finding is that sometimes the companies do not know which type of skills they need for their planned project, those who are frequently confused on the applicant side, as well.

The studies compare the positions from different aspects, and also the advertising companies (type, size, place, etc.).

text1, text2

They shared the code that they used to gain the data (website scratching and other).

One of their well organized graphs is clustering the skills and positions.