Background
Data monetization is the process of using data to increase revenue.
Why data monetization is essential for your organization ?

• Optimizes use of data.
• Extracts more, and better insights for you, your customers, and partners
• Streamlines decision-making & planning
• Increases operational productivity and efficiency
• Improves understanding of customers and customer experience

Methods
How well do you truly know your customers? 
Truly knowing your customers is about more than identity. It’s also about preferences, needs and the contextual knowledge of time and place. In the digital world, customers expect to be known as individuals with distinct preferences, not just as one member of a segment. Knowing your customer is powered 10 percent by your internal data and 90 percent by your ability to model the behavior of your customer

“KYC” refers to the steps taken by a financial institution (or business) to:

• Establish customer identity
• Understand the nature of the customer’s activities (primary goal is to satisfy that the source of the customer’s funds is legitimate)
• Assess money laundering risks associated with that customer for purposes of monitoring the customer’s activities

Results

Use Predictive Analytics to Keep Your Customers Well-Stocked, Predictive marketing analytics is a branch of advanced analytics that harnesses all that big data to predict future events or results. Machine-learning algorithms look at customers’ past buying patterns in order to predict their future actions. Using this strategy, you can identify potential threats that may affect their transactions journey and able to easily keep track of their customers’ transaction histories based on the date of transaction.

1. Give Personalized Recommendations for Individualized Experiences, Personalization is extremely important for customer loyalty. That’s because consumers are more likely to make a transaction if you give them something that they will actually like or are looking for. AI can help you discover exactly what your customers like quickly and easily. And then based on that, you can tailor the content you show them or the products you suggest for them.

Background

The project is focused to optimize the promotion cost-efficiency for customer membership by providing recommendations of which customer group optimal for certain offer types, as well as generate more revenue through customer behavior insights.

Data

Transaction data includes offers received, viewed, and completed by members as well as the time in hours needed to do the actions

Methods

The data is processed to perform a clustering model which produces customer groups based on their behavior.
Then, these groups can be interpreted in order to achieve targeted marketing and optimize the cost.

Cluster 0 is not optimal for BOGO offers, while cluster 1 is not optimal for discount offers. Cluster 2 could generate more revenue by providing offers that fit their behavior.

The Results

Before Optimization: Cost-to-Revenue Ratio: 2%

After Optimization: Cost-to-Revenue Ratio: 1.3% (35% more efficient)

Generate up to 16.7% more revenue (from the ‘not interested’ group)

The optimization performance can be improved through more data records and variables, thus, more granular behavior can be captured.

*cost refers to membership promotion cost

Background

“A picture speaks a thousand words” is one of the most commonly used phrases. But the graphics speak more than that. The visual representation of data, in the form of graphs, helps us gain actionable insights and make better data-driven decisions based on that data.

The network consists of 2 elements, nodes, and edges. Nodes represent objects of interest, while edges represent relationships between them. Nodes can be of the same or different categories in the network graph. Thus, the network can show the relationship between people, while at the same time the network can show the connection.

Data

The network is about the connections and relationships between different people, the nodes require (at a minimum) complete ID and Labels columns. The edges require Source, Target, and Weight.

Methods

Based on the mathematics of graph theory, graph algorithms use the relationships between nodes to infer the organization and dynamics of complex systems. Network scientists use these algorithms to uncover hidden information, test hypotheses, and make predictions about behavior.

Nodes

• Can have Labels to classify nodes
• Labels have native indexes

Relationships

• Relate nodes by type and direction

Properties

• Attributes of Nodes & Relationships
• Stored as Name/Value pairs
• Can have indices and composite indexes

Results

Many businesses work with graphs. Some examples are:

• Telecom operators operate fixed or mobile networks which can be modeled as graphs.
• Telecom customers talk to one another and these relationships form graphs.

Why is it important now?

Graph analytics is important due to the expected market growth. According to a recent graph analytics market report, the graph analytics market size was ~$600 million in 2019, and it is expected to reach ~$2.5 billion by 2024, at a Compound Annual Growth Rate (CAGR) of 34% during the forecast period.

Background

Most companies have big data. Big data is a collection of data that is very large in volume but grows exponentially over time. Because you have a lot of data, you most likely don’t always know what’s happening with it and you may not know when your data behaving in a way it shouldn’t. Artificial intelligence (AI) anomaly detection will help your business identify issues around the clock faster and more reliably, increasing efficiency and driving performance.

Data

An automated anomaly detection system uses machine learning to establish the normal patterns of time series metrics. Anomaly detection can detect attacks on a  network that deviate from normal activities based on statistical probability. Statistical anomaly detection does not have an intelligent learning model which might cause a false alarm to have a high detection rate.

Methods

Here we will look at detecting anomalies with time series forecasting. Time series is any data related to time (daily, hourly, monthly, etc.). For example, daily store revenue is time-series data at the day level. Many use cases like demand estimation, sales forecasting are common time series forecasting problems that can be solved by algorithms like SARIMA, LSTM, Holtwinters, etc. Time series forecasting helps us prepare for future needs by forecasting them with current data. Once we have an estimate, we can use that data to detect anomalies by comparing them to actual ones.

Results

The challenges for anomaly detection is selecting the appropriate anomaly detection method for each metric, tuning it, and constantly re-evaluating that choice in case the metric itself switched to a new normal behavior and looking for unusual data points by manually examining every metric, however, is impractical for more than a few dozen metrics. – all without human interaction.

An anomaly detection solution must be sophisticated and comprehensive enough to compute not only one outcome, but also tell the entire story and point to the root cause with considerably more benefits and faster time.