It’s supposed to be an ML python interview where Ill be asked to take a dataset and train a model on it. Has anyone done this interview before? Im curious about the type of data (text/image/fraud) that Ill be given. #tech #mle #stripe
Just posted the same Q. When is your interview?
How was the interview? What did they ask?
They gave me a dataset and asked to build a classification model. Practice on any dataset from kaggle, it's an easy round
How'd you proceed? Sklearn? Did you use pipelines? or just raw dogged it? Did you end up using LR / Tree based? Since LR might require standard scaler etc.. ``` df = ... clean_df = pre_processing(df) train_df, test_df = train_test_split(clean_df, ...) model = LogisticRegression.fit(train_df) preds = model.predict(test_df) print(pr_auc(preds, test_df["y"])) ```
I haven't done anything for MLE, but I did do an assessment for an analyst role there. The data they gave me was a mess, and they offered zero clarification on column definitions for ambiguous column names, (I was told that's just what to expect with some data you will see at Stripe). Be prepared to have to do cleanup and make a lot of your own assumptions.