The Secret Sauce behind Data Driven Giants

Analysts have estimated that one third of Amazon’s sales come via their recommendation system.

Where do you think they got this recommendation system??

They created it themselves.

COMPANIES THAT DECIDED TO BE DATA-DRIVEN FIVE YEARS AGO had to solve three major issues to succeed and become monsters:

1. Technoslavia: there are so many tools to equip data teams.

From visualization to data preparation, from the creation of machine learning models to putting these models in production -- all of this generates a lot of plumbing to connect all the parts.

2. Collaboration:

These pioneers broke the CLIENT-PROVIDER RELATIONSHIP that exists INSIDE companies between the business, IT and analytics teams. They found a way to create a predictive application “dream team” with all the key players focused on building the best product possible. This means that even non-technical team members are involved from the start in the process of machine learning.

These collaborative teams will usually include DATA SCIENTISTS, who are hard to find and hard to hire. You don’t want them to spend too much time on data cleansing, plumbing, paperwork or production issues -- you want them focused, along with the rest of the team, on building models that generate value for the company.

Keep your Data Scientists Happy here :)

3. Production:

To get any value out of this application, the predictions need to be used by the business on a daily basis. This is what we mean by production -- the live, automated applications of predictions that trigger actions. Examples could include recommending a product to a user, or stopping a likely fraudulent transaction, or even stopping a machine before it fails.

The companies that emphasize production aren’t content with simple analysis; they put that analysis to work. Think back to last decade: the companies that emerged as winners in the digital world were the ones best at shipping and releasing -- something they achieved by actually creating a culture around that!

How can you become more like Amazon, whatever your sector?

These three reasons are the reasons for the creation of Dataiku Data Science Studio (DSS) three years ago. Our goal is to allow any company that has data and wants to create business value from it, to have those three issues solved and be the best at creating their own data-driven applications.

Dataiku DSS is a collaborative platform on top of open source technologies that allows both clickers and coders to work together and understand each other to achieve common strategic goals :

1. Connect to any type of data source.

2. Prepare your data with our homemade big data transformation engine, which has more than 80 processors that makes data preparation easier. You will be able to code in SQL, Hive, Pig, Impala or Python too if needed.

Acquire, prepare, filter, join, and copy your data with visual components.

Use your favorite (big data) programming languages to add arbitrary custom logic.

3. Create Machine Learning models, both for clustering and prediction, and prototype quickly in our visual interface. You Data Scientists out there will be able to code in R and Python, using your favorite libraries and notebooks.

4. Visualize the results, share them with the rest of the business, add some notes and limit the exchange of documentation. The aim is to disrupt the supplier/client relationship I talked about before.

5. Put your project in production: create an API culture and involve the product team. In the following decade, the companies that will win will be the companies that excel at putting things in production and using advanced automation of workflows to expose your machine learning models via APIs.

These are big problems that are still not solved in many companies.

We are 60 people based in SF, London, NY and Paris.

The core concept of our product is to have a unified data science platform where you can have data management capabilities and machine learning capabilities together in the same software.

We have seen many companies starting data teams either from scratch or reinventing themselves through data. Although the companies that are innovating with data science today tend to be the largest companies, in the next five years innovation will come from companies of all sizes.

Our vision is to enable anybody to CREATE his or her own value using data -- to use two pieces of data, combine them together, test an idea, create a segment and test it.

We are talking about taking control over your DATA DESTINY instead of following others.

More than 100 clients between Europe and the US have already adopted this collaborative way of working for good. Various large e-commerce companies have created their own recommendation systems, homepage and newsletter personalization, and churn prediciton models. Big industrial players have done predictive maintenance, logistical optimization, and marketing campaign optimization. International banks and insurance companies have built models for fraud detection, life moment detection, and so on.

That’s what Dataiku DSS is all about: bringing data, technology, and people – no matter their level of expertise or skill set - together one predictive analytics solution at a time.

YOU WANT TO SCHEDULE A DEMO FOR YOU AND YOUR TEAM? CLICK HERE!