1. Interactive Dashboards (Excel)

Mix Sales data was provided- Sales of following products (Lipstick, Eyeliner, Mascara, Lip gloss, foundation) by sales person. There were 6 columns in the Sheet.

· Name of the Sales person

· The date/year of sale

· The product that was sold

· How many units were sold

· Amount collected by sales person with respect to the product and no of units sold

· Location of the sale

In order to get Quick Insights from the data and to better communicate the insights to the team data visualisation was necessary.

Insights –

· Top 5 Sales person – Rewards were given to the top sales person according to the Location and year

· Product Contribution – Year wise and location wise the sales of respective products were observed

· Sales trend – In a particular year the monthly sale of products can observed.

· For example – In year 2004 in Nashik location we can observe that the sale spiked in April and September and also the Top 5 sales person who achieved those sales.

2. Customer Segmentation of Mall Data using K means Algorithm and python Implementation

It is crucial to analyse the massive amounts of data that our environment frequently produces. The corporate strategy must be adjusted to the present circumstances in this modern period of innovation, where everyone is striving to outperform one another. Today's business depends on new ideas because so many potential customers are unsure about what to buy and what not to buy. Businesses can't assess their target market on their own. In order to improve decision-making, machine learning is used to find hidden patterns in data using a variety of techniques. Comparing data points from several groups is a key component of the machine learning process known as clustering. Applications include image processing, pattern identification, market research, medical data, search engine optimization, and others are among them. Our project is about customer segmentation, which is a subset of market research. The division of consumers into groups according to shared traits is known as customer segmentation. Businesses must categorise their customers into groups nowadays based on factors like age, gender, location, and other factors. This enables companies to concentrate on a select group of customers who are most likely to buy their products. If they can successfully implement machine learning to enhance their operations, they will have a competitive advantage over their rivals. The major objective of this project is to use the K-means algorithm to categorise clients according to their attributes.

By targeting specific consumer groups using a customer segmentation plan, businesses may allocate marketing resources more effectively, increasing the chance of cross- and up-selling. When businesses give personalised messages to a set of clients as part of a marketing mix suited to their needs, it is simpler for them to develop novel offers to persuade them to spend more money. By enhancing customer service, consumer segmentation may aid in increasing customer loyalty and retention. Marketing materials that use customer segmentation are more valued and appreciated by the consumer who receives them because they are more individualised than impersonal brand messages that disregard purchase history or any form of customer relationship.

Clustering has been shown to be advantageous for customer segmentation. Finding clusters in un labelled datasets via clustering is a form of unsupervised learning. K-means, hierarchical clustering, DBSCAN clustering, and more clustering algorithms are available. This work's main goal is to use a data mining technique to identify consumer groups by employing the K-means clustering method to divide the data.

Customer Mall Data was used from Kaggle website and the k means algorithm was implemented

Working of the Algorithm – K means is an iterative Algorithm.

Step 1: Select the number of K to determine the number of clusters (The K value is a intuition number at initial stage after using the Elbow graph we get the optimum no of clusters for segmentation)

Step 2: At random, select K locations or centroids. (It's possible that it's not the same as the incoming dataset.)

Step 3: Form the set K clusters by assigning each data point to the centroid that is closest to it.

Step 4: Calculate the WCSS and move the centroid of each cluster.

Step 5: Reverse the previous three steps, reassigning each datapoint to the cluster's new closest centroid.

Step-6: Go to step-4 if there is a reassignment; otherwise, go to FINISH.

Step 7: The model is now complete.

 Elbow Graph to Determine the optimum no of clusters

The Final Output –

· Insights : The output image is clearly showing the five different clusters with different colours.

· The Clusters are formed between two parameters of the data set, Annual Income and Spending Score

· Red Cluster – shows the customers with average salary and average spending so we can understand that despite their financial limits they have a desire to buy items.

· Violet Cluster – Low Income group with low spending score, these type of Customers are sensible as they plan according to their budget.

· Blue cluster – low Income group with high spending score, these type of customer’s can be categorized as careless and also profitable for the mall as they can response to offline touchpoints and can be influenced easily for on spot buying.

· Yellow cluster – high income group with high spending score can be categorised as target and these customers can be the most profitable customer for the mall owner

· Green cluster – high income group with low spending score, these type of customer are careful about their Spendings.