Management Analytics

INFO9026
Siamak Khayyati
HEC Liège - University of Liège
2024-2025
Some Information
• Location: N1a 59 (0/59)
• Time: Thursdays 16:15-19:15
• Email: [email protected]
• Evaluation:
• Project (5/20)
• Groups of up to 5 students
• Exam (15/20)
• Bonus exercises

2
Tentative Outline
• Introduction • K-nearest neighbors method
• What’s machine learning? • Neural networks
• Data preparation • Decision trees
• Linear and logistic • Random forests
regression • Clustering
• Python basics • Data visualization
• Project introduction • Sample exam
• Time series • Wrap up

3
Outline
• Predictive analytics
• What is machine learning?
• Data preparation with Excel
• Installing Jupyter notebooks

4
Predictive
analytics

5
Predictive analytics
• Business analytics
• Knowing what has
happened.
• Understanding
patterns in
performance
• Predicting what is
most likely to happen
• Identifying the best
set of decisions

6
What is machine learning?
Machine Learning
Branch of Artificial Intelligence Study of algorithms and statistical models allowing computers to
automatically learn from the data.

7
What is machine learning?
Association Analysis 35% of the purchases on Amazon
are the result of their
recommender system [McKinsey
2018]

8
What is machine learning?
Recommendations are responsible for 70% of the time people spend watching
videos on YouTube. https://www.cnet.com/news/youtube-ces-2018-neal-mohan/

9
What is machine learning?
• Association Analysis
• Provide Superior Customer Experience benefit from experience of others
• Increase conversion rates convert customers to a certain product/service
• Anticipate customer behavior
• Target marketing campaigns
• Co-branding activations

10
What is machine learning?
Segmentation
Manual segmentation
The analyst defines the groups manually
Example : group by income

Heuristic segmentation : RFM

Recency – How recently did the customer
purchase?
Frequency – How often do they purchase?
Monetary Value – How much do they spend?
Cluster analysis: (example : K-means algorithm)
Automatic segmentation

11
What is machine learning?
• Segmentation
• Provide Superior Customer Experience
• Segments based on individual preferences
• Develop Effective Retention Strategies
• Segments based on purchasing data and behaviour
• Better Ad Targeting
• Segments based on the lifecycle of a customer or on social channels they
frequent

12
 What is machine learning?
Predictive methods
Example: Predict customer response, e.g. ordering a product online with a promotional code
their response on ad campaign (did they
1. Build a prediction rule using a training dataset (historical records) buy smtg ?)

Age Gender TOTSales GoldMember Country Children WebOrder Response

23 F 1256 Silver BE 0 8 Yes
36 F 235 Silver LUX 2 2 No
54 M 32 No FR 1 0 Yes
19 F 601 Gold BE 0 3 Yes
22 M 412 No BE 0 6 No

2. Predict new cases on the basis of the built prediction model

Age Gender TOTSales GoldMember Country Children WebOrder Response
find rules to help you find these question marks (predict
response) using case values where we actually have the42 F 456 No BE 3 2 ????
responses)
31 M 879 Gold LUX 0 4 ????

13
What is machine learning?
• Predictive Methods
• Predict consumer behavior
• Predict campaign performance taux de perte de clientèle

• Predict attrition/ churn to launch retention programs

• Predict customer preferences
• Target the most profitable products to customers most likely to buy
them

14
What is machine learning?
• Applications
• Predict churn
• Mobile phone and utilities companies use Machine Learning
to predict ‘churn’, i.e. when a customer leaves their company
to get their phone/broadband from another provider.
• Fraud detection
• Identify customer fraud and risk behavior patterns across
multiple lines of data. Score risky transactions that need
further investigations.

15
What is machine learning?
• Applications
• CRM : Identify prospective customers, social network,
recommendation
• Supply chain management: sales forecasting and
production needs, diagnosis of machine faults,…
• Finance: Credit risk (the chance that a borrower does
not repay a loan or fulfill a loan obligation)
• Medicine : monitor intensive care patients

16
What is machine learning?
• Applications
• Image detection
• Optical character recognition (OCR)

17
Data preparation
with Excel

18
Exercise 1
• Using absolute and relative
references
• In a new Excel workbook,
compute efficiently a table
with the final values of an
investment on the basis of
the formula 𝐶 × 1 + 𝑟 𝑡 for
• 𝐶 = 1000
• 𝑟 = 0.01 to 0.1 by steps of 0.005
• 𝑡 = 1 to 20 by steps of 1

19
Exercise 2
• The file Ecercise_2.xlsx contains for each month of a given
year, the price for one gas liter and the number of liters
consumed by a given family for each month.
• Compute, in column E, for each month, the gas cost
• Compute, in E15, the total cost for the year.
• Let’s assume that the family negotiates a discount of 1,5% on
the gas price for the year. Compute the reduced price for each
month in column H, and the reduced total cost in column
H15.

20
Exercise 3
• The second sheet in Ecercise_2.xlsx only contains the information
about the consumed gas liters.
• Compute manually, in C15, the average number of consumed
liters. Compare with the predefined function “average” in B15.
• Compute manually, in C16, the median and compare with the
predefined function “median” in B16.
• In which case is it preferable to use the median instead of the
mean?
• Compute manually, in B19, the standard deviation. Compare, in
C19, with the predefined function “stdev.p”.

21
Exercise 4
• Importing data
• Import the file Exercise_4.txt
• Compute a new column with the Total retail price per unit

22
Exercise 5
• Histograms, samples and joining tables
• Use the file order_complete.xlsx and Plot a histogram of the Total Retail
Price

Histogram
60000

50000

Histogram of the TRP: 40000

X-axis : TRP in bins 30000

Y-axis : the height of each rectangle
20000
corresponds to the number of values
10000
in column J falling in that bin
0

23
Exercise 6
• In the file order_complete.xlsx, in the sheet customer_dim Replace the
term Orion Club Gold members low activity by low and Orion Club
Gold members high activity by high
• How can you select a random sample of 1000 customers ?
• Complete the sheet order_fact with four columns with information
coming from the other sheets (not using copy and paste but based on
the product id, the customer id and street id)
• the product category and product name
• the customer type
• the country

24
Exercise 7

• Use the file order_complete.xls

A. Show all the transactions for the product category Clothes
B. Show all the transactions for the product category Shoes in Australia

25
Python and Jupyter Notebook
• We will use the programming language Python for using
implementations of machine learning algorithms
• To ease the introduction to Python, Jupyter Notebook is
used as an integrated development environment
• As a Python distribution, we will use Anaconda
• The following slides explain the required installation steps

26
Python and Jupyter Notebook
• You can find Anaconda installers https://repo.anaconda.com/archive/
• Select the correct graphical installer for your operating system and download the executable
• Select and download the Anaconda3-2023.03-1 version
• Open and run the executable file, then click Next
• Select an installation location and click Next, make sure to have no spaces in path
• If there is the option, check the Add Anaconda to my PATH environment variable checkbox!
• Click Install

27
Python and Jupyter Notebook
• Go to Windows start menu and look for ”Jupyter
Notebook” and click on it

28
Python and Jupyter Notebook
• The notebook interface will appear in a new
browser window or tab

29
Python and Jupyter Notebook
• Click on New > Python 3

• A new tab/window will open with a new Jupyter notebook.

30
Python and Jupyter Notebook
• Write following in the first line: print(”Hello, world!”)
• Click on Shift + Enter to run the code

31
Python and Jupyter Notebook
• Click on File > Rename to rename the notebook file

32
Python and Jupyter Notebook
• Click on save icon to save changes

33
Python and Jupyter Notebook
• After closing the windows, you can navigate in the notebook
interface and click on the file name to open the notebook

• Installing packages:
• In Anaconda Prompt:
• pip install pandas
• pip install numpy
• pip install matplotlib
• …
34
Links
• https://python-
course.eu/books/bernd_klein_python_and_machine_learning_
a4.pdf

35