PREDICTIVE
MODELING
AUTOMATION
Platform
Powered by AI
CREATE FORECASTING MODELS online, fast and easy
concept
A Solution that will change the industry

ANTAVIRA helps to build forecasting models faster and easier. It reduces the time spent on manual work and result computation.

ANTAVIRA allows you to make up to 1,000 statistical models simultaneously for different targets using a large number of predictors without repeating the same actions.

ANTAVIRA combines the tools used at each modeling step from sampling to models staking. Each tool can be used in conjunction with others, or separately.

ANTAVIRA is suitable for data-scientists of any level. Every professional can find useful functionality from the system.

Benefits
Designed to become an undouptable necessity
EARN MORE MONEY

ANTAVIRA reduces building time of a good quality model GINI>0.5 from 15 working days to 2 working days.

Make more models at higher quality to get better paid

SAVE TIME

ANTAVIRA automates modeling process from samples preparation to models staking. No need to repeat set of actions unless a good model is created.

Decrease time of manual work and results computation by 60%

MAKE MORE MODELS

ANTAVIRA creates a large number of statistical models for different target variables using a large number of predictors simultaneously.

Launch more modeling experiments to get better result

INCREASE QUALITY

ANTAVIRA maximizes prediction power on each step of modeling and controls over-fitting by using included AI mode.

Increase prediction power, get better AUC, GINI, KS and avoid over-fitting

Tools
Features and Functions
SAMPLE PREPARATION
Step №1
MODELING
Step №2
ENSEMBLER MODELING
Step №3
Each module can be used separately as well as in modeling chain
Each module can be used separately as well as in modeling chain
Each module can be used separately as well as in modeling chain
  • Assembler
    Assembles sample [variables] with sample [targets] and makes samples [variables+targets]. Number of samples [variables+targets] is equal to number of targets
  • Splitter
    Splits samples [variables+targets] into t raining and testing samples. Where are 3 methods to split: Percent ratio; Timeline; Cross-validation
  • Binning
    Bins every variable in every training sample using AI algorithm to maximize prediction power and avoid overfitting
  • Correlation
    Measures correlation V-Kramer and chooses most predicable variable in every training sample using AI algorithm
  • Modeling
    Creates set of models snd picks best one as a final result for each sample using AI algorithm
  • Score Calculation
    Calculates scores for chosen models for any sample provided or samles from modeling chain
  • Model Stacking
    Chooses best ensembling target, creates setof meta-models basedon models scores and picks best oneas a final result by using our innovative methods and AI algorithm
  • Final Score Calculation
    Calculates final scores of any chosen meta-models for any sample provided or samples from modeling chain
  • Sample preparation
  • Modeling
  • Ensemble modeling
Price

We provide free access
to Beta version of Antavira

in return on your comments
after 15 days of use

Experience
Reputation that can be trusted
5
years
233
clients
5+
countries
45 000
variables
32 000
models
9 000 000
credit decision

ANTAVIRA predictive modeling automation platform is a project of SCORISTA – the world’s first credit risk management service for non-bank lenders.

SCORISTA provides not only the credit score, but instant credit decision - to "Approve" or "Deny" the credit with detailed explanation in each case. SCORISTA is the only credit risk management company with a money-back guarantee program.

SCORISTA was founded in 2014, and currently produces 9,000,000+ credit decisions for 233+ clients in 5 countries: Russian, China, Spain, Ukraine, and Kazakhstan. SCORISTA will expand to the U.S market in early 2020.

Innotribe Challenge, 2017
InnoVex, 2017
Golden Ruble Award, 2017
Bank Review, 2017
Startup of the year, 2017
Sibos, 2017
Tech Tour, 2017