Data Analytics
Platform
Facilitating and fasttracking your business AI with machine learning-driven data formating, blending and model build for predictive analytics
Take the business AI fasttrack
Our Data Analytics Platform help you navigate from big data to smart data, so you can take the right, growth-driving decisions for your business. The Data Analytics Platform is ideal if you are about to build models for predictive analytics or want to benchmarks the models you already have, especially within:
- customer segmentation and improved CEM
- churn/retention
- cross-sell and up-sell
- newbizz/improved conversion rates
- risk management
- etc.
The platform optimises your ressources because our machine learning-based tools significantly shortens the process compared to having to do it all by yourself; you get the right-sized data pool to build valuable models with our comprehensive data lake which gives you the perfect foundation to identify relevant trends and triggers together with your internal customer data to improve your existing customer flows
We ease your data science with the Data Analytics Platform with our machine learning-driven tool so:
- structure your internal data and make them ready for business AI
- blend it with the variables in our Nordic data lake
- identify the important variables for your model (churn, segmentation, upsell, newbizz etc.)
- build models for predictive analytics and help you implement them
Navigate this site
Making data ready for business AI
Ahead of any model building for predictive analytics it is key to collect data and make it ready. It is not a simple process as it demands the localisation, selection, blending, structuring, and formatting of data
We cut short this task and make it more agile with our formatting algorithms; a tool to format and structure your data quickly to make them ready for the business AI model building. You internal data consists of different data sources and components (degrees of data) and we handle all types of data
Our model data formater tool covers:
- detecting and handling different variable types
- removing highly correlated variables
- data imputation on missing data
- outlier detection and removal
- data Standardisation
The data formating process
keyboard_arrow_up
keyboard_arrow_down
Our data blending engine:
- matching & geo-coding algorithms
- data formatting and data computation:
- shuffle rows
- remove empty variables and variables with more than 20% missing values
- change character/string variables to categorical
- classify variables with 25 unique levels or less as categorical, and as continuous otherwise
- add variable suffixes identifying types
- remove sequential variables (e.g. 1,2,3,4..)
- remove constant variables and variables with near zero variance
- remove duplicated variables
- remove rows with more than 40% missing values
- validation checker
- dimension controls
- principal components
- correlation analysis splits
By automating the data formating process, you satisfy your data scientists by removing a task that does not add value – allowing them to speed up the time to build and execute new models. And not only does this speed up your process to build a sufficient data universe for your AI; it also enriches your internal customer data, giving full insights that ensure your models view your customers as holistic human beings; this is what we call getting the full 360° view on your customers
Data blending: your internal customer data blended and ranked with 3rd party data
The next step in the process to get ready for predictive analytics model building is to blend your internal customer data with the external 3rd party data in our Nordic data lake. The data blending ensures you a large, relevant data pool; necessary to build valuable models
With our Nordic data lake you get easy access to more than 800 variables already structured and ready to use. Our data blending engine is the tool that via complex matching and data formatting algorithms, matches and blends your internal customer data to the data components in our data lake in a controlled environment, and rank their significance for the model you are about to build; ready for the model development stage
Optimise your data pool with our Nordic data lake
Our Nordic data lake consists of 800+ 3rd party variables. By signing up for our data lake, you can access all of our variables 24/7 and it is software agnostic. This cuts significant time for your build of a data repository sufficient for your AI and model building needs, as well as all the trouble of structuring, validating, documenting, and updating of 3rd party data. And naturally, you get easy access to our expertise within register data, GDPR compliance, and data science
Our data lake is the centralised repository of all our data components, covering
- Data (Nordic):
- raw data
- structured data
- modelled data
- Fully documented
Identifying the significant variables
The blending process also involves identifying the relevant and most significant variables for the model you are about to build; churn, cross-sell, up-sell, newbizz, etc. Through our predictive machine learning engines, we identify the variables that are significant for your business; what parameter sets your customer segments apart from each other?
The result is a report, which shows based on your target goal, what variables perform and are the more significant (both from your internal data and from our 3rd party data); and which of our machine learning engines is best suited to service your type of analysis
The report gives you an overview of your own data along with our 3rd party data; listing the significant and beneficial variables for your model building process, saving you the time and trouble of a manual process
These most significant variables can then be accessible and enriched on to each and every customer record; allowing you to take these variables into further analysis and modelling steps in-house or via our data science services
Benchmark your existing models
Our data blending engine can also be used to optimise the data usage and selection within your internal model building process by identifying the predictive nature and auto-selecting the relevant data for further investigation and modelling. By accessing our blending environment, we enable you to benchmark your internal models and optmise them by adding the relevant 3rd party data
Predict: model building with machine learning
Predict is an AI tool that blends the different data into predictive models driven by machine learning protocols. It helps you develop better models and speed up the time-to-market of your prediction tool; we have all the structure and framework around the non-direct modelling tasks and we generate the models in the self-service environment. You can call it automated AI if you will. So, even if your internal data is not ready for AI, we can help you build models that will give you a better understanding of your customers and your market - and you have the benefit of our in-built security as our data are already discretionised and structured to comply with GDPR. In short, we help you grow and optimise your data assets
Our Data Analytics Platform generates deployable models and outputs that score your base after the desired targeted goals; coupled with the ability to then assign recommendation action-codes. Not only does it enhance your understanding of your existing customers; it also ensures that you can twin your internal models to the market - and easily adjust them over time as needed
Get started with business AI, even if your internal data are not ready
Predict works with a minimum of your internal customer data so even if your data are not ready for AI yet, we can help you build models ensuring you a better understanding of your customers and your market. And you get the security of our built-in privacy as the data in our data lake is already GDPR-compliant, discretioned and structured
Access points
Automised workflow
FTPs or SFTP server
3rd party plugins
