Segmenting the market based on shopping patterns using k$-$means

$1\mathrm{.}0$ Description:

Students will apply unsupervised learning techniques to segment the market based on

customer shopping habits using k$-$means algorithm. Students will use a file named

sales.csv$.$ The file contains the sales details of a variety of tops from several retail

clothing stores, where each datapoint is defined by $6$ attributes. To complete this

problem students can refer to the section of k$-$means algorithm and the last section on

chapter $7$ of the textbook and the handout.

$2\mathrm{.}0$ Algorithm $($The main steps$)$:

$1-$ Import the required packages from csv$,$ numby, matplotlib.pyplot, sklearn.cluster

$($kmeans$),$ sklearn $($metrics$).$

$2-$ Load the data from the input file $($sales$.$csv$).$ Since it's a CSV file, you can use the csv reader

in Python to read the data from this file and convert it into a NumPy array $($X$).$

$3-$ Define the number of clusters $($num$\_$clusters$)$ before applying the K$-$Means algorithm.

Here, you can assume that data lies on $9$ different clusters.

$4-$ Create the KMeans object $($kmeans$)$ using the initialization parameters. Be carful to use the

right parameters.

$5-$ Train the kmeans model with the input data.

$6-$ Extract and print the centers of the $9$ clusters $($cluster$\_$centers$).$

$7-$ Since you are dealing with six$-$dimensional data and in order to visualize the data, you can

take a two$-$dimensional data formed using second and third dimensions. So$,$ create two

subsets from data and centers of only second and third attributes $($X$\_2$d &

cluster$\_$centers$\_2$d$).$

$8-$ To visualize the boundaries, you need to create a grid of points and evaluate the model on

all those points. So$,$ you can start by using the step size of $0\mathrm{.}01($step$\_$size $=0\mathrm{.}01).$

$9-$ Define the grid of points and ensure that you are covering all the values in the input data.

$10-$ Predict the outputs for all the points on the grid using the trained K$-$Means model.

$11-$ Plot all output values and color each region.

$12-$ Overlay input data points on top of these colored regions.

$13-$ Plot the centers of the clusters obtained using the K$-$Means algorithm.

$3\mathrm{.}0$ Sample Output:

To see sample output for same related example, students can refer to the section of the k$-$

means algorithm and the last section of the same title on chapter $7$ of the textbook and the

handout.

$2$

$4\mathrm{.}0$ Implementation to be completed:

# Import the required packages

# Load the data from the input file and convert it to numpy array

input$\_$file $=$ 'sales.csv$\text{'}$

file$\_$reader $=$ csv$.$reader$($open$($input$\_$file, $\text{'}$r$\text{'}),$ delimiter$=\text{'},\text{'})$

X $=[]$

for count, row in enumerate$($file$\_$reader$)$:

if not count:

names $=$ row$[1$:$]$

continue

X$.$append$([$float$($x$)$ for x in row$[1$:$]])$

X $=$ np$.$array$($X$)$

# Define the number of clusters, create the KMeans object $($kmeans$)$ using the initialization

# parameters, and train the kmeans model with the input data.

num$\_$clusters $=9$

# Extract and print the centers of the $9$ clusters.

# Create two subsets from data and centers of only second and third attributes.

X$\_2$d $=$ X $[$:$,1$:$3]$

cluster$\_$centers$\_2$d $=$ cluster$\_$centers $[$:$,1$:$3]$

# Step size of the mesh

step$\_$size $=0\mathrm{.}01$

# Define the grid of points using the sub$-$dataset X$\_2$d$.$

# Predict the outputs for all the points on the grid using the trained kmeans model.

$3$

# Plot all output values and color each region.

output $=$ output.reshape$($x$\_$vals.shape$)$

plt$.$figure$()$

plt$.$clf$()$

plt$.$imshow$($output$,$ interpolation$=$'nearest', extent$=($x$\_$vals.min$(),$ x$\_$vals.max$(),$

y$\_$vals.min$(),$ y$\_$vals.max$()),$ cmap$=$plt$.$cm$.$Paired, aspect$=$'auto', origin$=$'lower'$)$

# Overlay input data points on top of these colored regions.

# Plot the centers of the clusters obtained using the K$-$Means algorithm.