Quickstart

Prerequisites

Python 3.11 or 3.12
A Hollerith account
An API key — see Get your API key

Install the SDK

Hollerith is a private SDK, install with this command:

pip install https://hollerith.monarcha.ai/sdk/hollerith-0.1.2.dev157-py3-none-any.whl

Set your API key.

export HOLLERITH_API_KEY="hk_live_YOUR_KEY_HERE"
export HOLLERITH_BASE_URL="https://hollerith.monarcha.ai"

Confirm the install:

python -c "import hollerith; print(hollerith.__version__)"

0.1.2.dev157

Make your first prediction

This example uses the hotel booking sample from the console: 19 952 reservations, one labeled booking column reading cancelled or kept.

from hollerith import Hollerith
import pandas as pd

df = pd.read_csv("hotel-cancellations.csv")
split = df.pop("__hollerith_split")
train = df[split == "train"]
test = df[split == "test"]

model = Hollerith()
model.fit(train, target="booking")
preds = model.predict(test.drop(columns="booking"))
print(preds[:5])

['kept', 'cancelled', 'kept', 'kept', 'cancelled']

fit uploads the table and returns a context to predict against. predict reads it.

Measure the accuracy

Pass evaluate=True to fit and Hollerith scores itself on held-out rows.

model.fit(train, target="booking", evaluate=True)
print(model.evaluation_)

One metric comes back: accuracy for classification, RMSE for regression. Evaluating accuracy covers which one you get and how the method behind it is chosen.evaluate=True runs a second job and bills it separately.

The context is reusable. Predict against it as many times as you need without fitting again.

Swap in your own table

Point read_csv at your file and hold out your own test rows:

df = pd.read_csv("your_data.csv")
train = df.sample(frac=0.8, random_state=0)
test = df.drop(train.index)

model.fit(train, target="your_label_column")
preds = model.predict(test.drop(columns="your_label_column"))

Predict a number

Same shape, numeric target. Hollerith reads the task from the target column. This example uses the concrete-strength sample: 1 030 mixes, eight numeric columns, one strength_mpa value from 2.3 to 82.6.

df = pd.read_csv("concrete-strength.csv")
split = df.pop("__hollerith_split")
train = df[split == "train"]

model = Hollerith()
model.fit(train, target="strength_mpa", evaluate=True)
print(model.evaluation_)

Evaluation(RMSE=3.2, method=kfold, folds=5, rows=824)

RMSE is in the target’s own units, so 3.2 is 3.2 MPa against mixes spanning 2.3 to 82.6. A target of small whole numbers is read as classification unless you say otherwise. Pass task="regression" when that is wrong.

Get started

Guides

Reference

Account

Help

Prerequisites

Install the SDK

Make your first prediction

Measure the accuracy

Swap in your own table

Predict a number

Next steps

Evaluating accuracy

Python SDK

​Prerequisites

Install the SDK

Make your first prediction

Measure the accuracy

Swap in your own table

​Predict a number

​Next steps

Evaluating accuracy

Python SDK

Prerequisites

Predict a number

Next steps