This application isn’t interactive (sorry!), but we’ll explain
why. The other apps shown here use ScienceOps to process
input and make predictions in real-time. But ScienceOps can
also be used in “batch mode,” which is what this example
A data science team has built a model to rank which leads the sales team should contact first. They deploy this model to ScienceOps, where it is “called” every night by their CRM system (think Salesforce). Every morning when the sales team comes in to work, Salesforce shows the results, a rank ordered list of which leads are the hottest.
83 lm.fit(df[features], df.converted) 84 85 rf = RandomForestClassifier() 86 rf.fit(df[features], df.converted) 87 ...
134 probs <- predict(logit, newdata=df, type="response") 135 grades <- cut(probs, 5, labels=c("F","D","C","B","A"), 136 ordered_result=TRUE) 137 lead_quality <- table(grades) 138 ...