What is data science & predictive analysis?
Highly successful businesses know that they can no longer depend solely on their product or service to develop; they must leverage their data to a better understanding of their customers as well as learn from the collective experiences of their organizations to stay competitive.
Data Science & Predictive Analytics for Business Professionals offers individuals the skills needed to efficiently collect and manage Big Data and perform data-driven discovery and prediction, extracting value as well as competitive intelligence for their organizations.
Why data science & predictive analysis is growing?
There’s no doubt that predictive analytics has been all the rage for the last few years—and for good reason. The new algorithm-based discipline has empowered us to bring about insights that provide details on the probability of a given outcome, as well as helping us bring about the ones that are favorable.
Use cases of data science and predictive modeling comes in the form of trip-planning tools, where customers can set locations, dates, miles program memberships, hotel needs, and other aspects that affect travel details. These products mark an evolution toward user-friendly data science, with tools that essentially turn the customer into the data scientist, allowing them to create their own models to return desired results.
Why every business is adopting this technology?
You must have heard the famous quote, ‘If you don’t know where you are going, then any road will get you there.’ However in a data-driven world where the contention among organizations to get dynamic is getting close with each passing day. Every business organization is peddling up to capture all the data that streams into their businesses and then apply analytics to get significant value from it. Big data analytics is helping organizations harness their data and using it to identify new opportunities. This is helping them to make smarter business moves, more efficient operations, higher profits, and happier customers. At such times, when businesses are using basic analytics to uncover insights and trends, big data analytics has also brought a world of opportunities to the table.
The future of data will be witness continued exponential growth. Thereby, businesses must learn to deal with timely actions from the insights these data sets create. Increasingly, real-time actions are required but this really depends on the business type and whether what they do is time-critical. However, real-time or not, data is changing how we do business, and companies that successfully combine their domain expertise with data science will out-compete their rivals every time.
Predictive analytics will allow a continuous analysis of customer data, while Machine Learning capabilities will provide the most relevant results and recommendations to users. Many predictive analytics solutions are being made available for on-cloud deployment, which is compatible with multiple e-commerce platforms. Not all customers will interact with an e-commerce Store in the same way. Every customer is unique and their online behaviour will differ based on individual tastes and preferences. Predictive analytics helps to assess different variable elements in customers’ behaviour. This will generate the desired engagement and responses from the customer, making their e-commerce experiences highly personalized.
Today, we have access to high volumes of data, coming from a variety of data sources. Thereby, every line of business can be optimized by implementing insights derived by big data analytics tools.
Common methodology for Problem Solving - CRISP-DM
Data Science Use Cases
Idea Cellular Churn Prediction
- All telecom companies across the world face one common problem i.e. which are the customers that they should be targeting to increase revenue
- Idea cellular Ltd had a similar problem
- They wanted to identity the customers who they could reach out to with some promotional offers/discounts before the subscribers churn or leave the company's network for the competition which could in turn affect their topline
- The problem at hand was a typical machine learning classification problem which could predict such subscribers who had the maximum probability to churn
- A model was proposed for each circle that idea operates in, a POC was conducted for a few circles to see if the hypothesis worked
- We followed industry standard CRISP-DM methodology to generate insights from the data beginning with problem formulation i.e. defining the problem in a mathematical form and proceeding with data exploration and finally model building
- The proposed solution was built using Logistic Regression optimized using gradient descent in SAS
- Solution involved building a custom model for each circle where Idea was a leader in terms of subscriber share or one amongst the top 2 leading companies of that circle
- Models for each circle were managed by a team of analysts and marketers who framed the promotional offers for subscribers
- We tested the models using standard A/B testing to see if our models were driving any business value
- On comparison between the subscriber groups were the model was applied vs the groups where the model was not applied, the uplift in revenue was approx. 2x thereby contributing a significant proportion to the companies topline
- Models for various circles worked efficiently and helped Idea retain/acquire the top position in some of them by retaining the customers that had a high propensity to churn
Maersk Hull Cleaning Prediction
- Maersk Tankers a subsidiary of Maersk Group very frequently had to cancel hull cleaning of their docked vessels due to a lot of False positives which indicated that the Hull required cleaning as per the fuel optimization experts but that was actually not the case
- This resulted in taking the vessel off hire which was a significant hit to the topline of the organization as the vessel had to be bought out of business to perform hull cleaning and was not operational due to this
- We found out that the stand alone reports generated by the fuel optimization team were not enough to gauge whether the hull required cleaning or not and that there was a lot of sensory data that the vessels were generating that was not being put to good use to make an intelligent decision regarding the same
- We decided to bring all the vessel related historical data together and build a comprehensive solution that could predict along with the intelligence bought by the fuel optimization team on whether the vessel should be bought to docks for hull cleaning or not
- The proposed solution used all the sensory data/logs from the vessel along with data from fuel optimization teams to build a random forest classifier that could provide additional intelligence on whether the vessel required hull cleaning or not
- We followed the industry standard CRISP-DM methodology to solve this problem where the clear understanding was to bring the false positive rates down and to make sure that every time the fuel optimization team and the model predicted that the hull cleaning was required, would eventually turn out to be true
- In case the model and the fuel experts were on the opposite ends, the business was asked to do a more thorough investigation before making an informed decision
- After the deployment of the solution, the results were monitored for a couple of quarters and it was found on comparing with the results of previous quarters that the model bought down the false positives by a margin of approx. 50%
- The model along with fuel optimization experts eventually was able to spot almost 100% of the times when the hull cleaning was required and when was it not required thereby helping the client achieve higher operational efficiencies and a better topline
Loveesh who is the Chief Data Scientist at Excelloite has been exceptional in his services, professionalism and technical acumen, this was one of the primary motives we worked with him and he proved that each and every time we faced a complex business challenge to drive revenues and cut the bottom line with the help of data. I wish him all the best for future endeavors and would sincerely like to thank him for being the spear head for solving complex data problems using machine learning and data.
Idea Cellular Ltd.
We were facing a very difficult problem of optimizing our fleet strategy to maximize revenues and didn't know how to do it till Team Excelloite came onboard. Their knowledge related to Data and algorithms combined with excellent understanding of our business helped us build a solution in less than a couple of months. Apart from building predictive algorithms that specifically solved our business challenges they also help us understand and visualize the data in a much more efficient way. We highly recommend Team Excelloite and would love to work with them again.