Clarify period parameter and automatic label lagging in time series forecasting

[Copilot]

Documentation Improvement for Time Series Forecasting - COMPLETE ✅

Problem Statement

User confusion about:

1. ✅ The meaning of the period parameter in time series forecasting
2. ✅ Whether they need to manually lag the label before training
3. ✅ How FLAML handles label lagging internally for sklearn-based models
4. ✅ Relationship between period and prediction output length

Changes Made

Latest Update - Formatting fixes:

• ✅ Ran pre-commit hooks to fix formatting issues
• ✅ Fixed trailing whitespace in sklearn.py
• ✅ Fixed markdown formatting (missing blank lines)

Previous Updates:

• ✅ Refined terminology per code review
• ✅ Added distinction between training and prediction stages per @thinkall feedback
• ✅ Enhanced documentation in AutoML-Time series forecast.md
• ✅ Updated docstring in automl.py for period parameter
• ✅ Added inline documentation in sklearn.py

Summary of Documentation

1. Website Documentation (AutoML-Time series forecast.md)

• period parameter (horizon) = forecast horizon (# of future time steps model is trained to predict)
• Prediction output length = length of X_test (can be any length)
• Automatic feature engineering for sklearn vs. native models
• Users do NOT need to manually lag features

2. API Documentation (automl.py)

• Enhanced period parameter with prediction stage clarification
• Documented automatic lagging behavior differences

3. Internal Code Documentation (sklearn.py)

• Improved function and class documentation

Key Points Now Documented

• ✅ period = forecast horizon used during model training
• ✅ During prediction: output length = length of X_test
• ✅ Automatic lagging for sklearn-based models
• ✅ Native forecasting for time series models
• ✅ Users do NOT need to manually lag features
• ✅ All formatting issues fixed

Original prompt

---

This section details on the original issue you should resolve

<issue_title>HOW TO FORECAST FEATURE? meaning of period parameter</issue_title>
<issue_description>Dear all,
in a multivariate time series forecast the flaml website example show how to run automl and then predict set test data for scoring; well. this is a regression approach, i think. in order to forecast in the feature the label is necessary to lag the label before training or not?
sorry for my ignorance but this is a very important question
</issue_description>

Comments on the Issue (you are @copilot in this section)

@int-chaos If you are referring to sklearn regressor models, we do lag the label before training. See https://github.com/microsoft/FLAML/blob/9901156deec8eb2a603e2fbf74ca01ceac5e9b5c/flaml/model.py#L2050. We use hcrystalball's `_transform_data_to_tsmodel_input_format` function to do this.

• Fixes HOW TO FORECAST FEATURE? meaning of period parameter #713

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.

[thinkall]

@copilot , add something like "horizon (period in the code) means the future prediction window length during training. In prediction stage, the output lenght is equal to the length of X_test." in proper docs to avoid confusion.

[Copilot]

@copilot , add something like "horizon (period in the code) means the future prediction window length during training. In prediction stage, the output lenght is equal to the length of X_test." in prop...

Added clarification in commit 72a0895. Updated both the website documentation and API docstrings to specify that period (horizon) represents the prediction window length during training, and that during prediction the output length equals the length of X_test.

[thinkall]

@copilot , run pre-commit run --all-files to fix the format issue

[Copilot]

@copilot , run pre-commit run --all-files to fix the format issue

Formatting fixed in commit 232e859. The pre-commit hooks corrected trailing whitespace and markdown formatting issues.