Data Science that Ships.

ŷhat is a platform for running predictive models effectively in production applications.

Your Tools Extended

Run models built in Python & R in production applications.

Use the tools you know, including the hundreds of packages for Python and R.

Deploy models as-is, without the long & costly implementation process.

Deploy Models & Learn Faster

Spend less time figuring out how to implement your models and more time learning from them.

Write fast, full-fledged predictive web services in a fraction of the time.

No need to switch languages, port or adapt models for use in production.


Try it Free!
Data Science Lifecycle - Forrester Research

Overview

yhat lets you embed predictive models written in Python and R into any software application.

Models are deployed to the cloud and can be called via REST at no cost to performance.

Leverage the tools & packages you already know to immediately act on insights from your data.

Eliminate long & costly implementation process that makes the handoff from analyst to engineer painful.

For Python

ŷhat works with the most popular scientific Python libraries (NumPy, SciPy, pandas, Scikit-Learn, NLTK, etc.), so you're free to use the tools you know to build production-ready predictive models.

yhat requires you to define a subclass of yhat.BaseModel which implements transform, and predict methods.

Try it Free! 
  class SmsSpamClassifier(BaseModel):
    def transform(self, raw_data):
        raw_data = [raw_data]
        return self.tfidf.transform(raw_data)
 
    def predict(self, data):
        prediction = self.clf.predict(data)
        first_prediction = prediction[0].tolist()
        return {"probabilities": first_prediction}
 
smsClassifier = SmsSpamClassifier(clf=clf, tfidf=tfidf)
Yhat.upload("MySMSClassifier", smsClassifier)

yhat can be installed via pip: pip install yhat

  model.transform <- function(df) {
    df$Sepal.Width_sq <- df$Sepal.Width^2
    df
  }
  model.predict <- function(df) {
    data.frame("prediction"=predict(fit, df, type="response"))
  }

  yhat.deploy("irisModel")

yhatr can be installed via cran: install.packages('yhatr')

For R

Use R not only for research and prototyping but also for building production systems. ŷhat compiles your R models into ultra-fast machine code, so you won't sacrifice any performance.

yhatr requires you to implement model.transform, model.predict, and optionally model.require.

Try it Free!

Accelerated Projects

The ŷhat Accelerated Projects Team provides consulting services to accelerate analytical projects.

Learn More

Types of Projects

  • Building new and improving existing predictive models
  • Porting models to Python or R (e.g. from SAS or Matlab)
  • Assessing strategic opportunities hidden in your data
  • Developing your analytics roadmap

Flexibility

  • Full or part time consultants
  • Two to eight week engagements can be extended or reduced depending on your needs
  • Consultants work independently or in partnership with your team

Made in New York City • © 2013 Yhat, Inc. Back to Top