Reference no: EM133675056

Programming AI for Business Analytics

Assessment - Skills-Building: Causal AI with Python

Assessment - Practical Python Causal Artificial Intelligence Coding

Your Task

In class, Follow the steps within the Google Colab notebook to complete a Python code.

Background
Imagine that you are a Business Analyst in a top consulting firm and that you have been tasked with the identification of factors that cause Australian macroeconomic changes.

This skills-building exercise is a workflow that has been designed to simulate this use case and consolidate your practical knowledge of Python, Google Colab, and machine-learning approaches for causal inference.

The Google Colab notebook for this assessment will be provided on the day of the assessment.

Assessment Instructions

Section #1: Machine Learning
1A: Installation & Module Imports

Step 1: Install Python libraries and import Python modules.
Run all the codes that will install and import the necessary Python libraries and models.

Step 2: Load and import macroeconomic data as a dataframe.
Load data: read csv file containing macroeconomic data. Create a suitable variable name for the dataframe.
Write the Python code that displays information about the dataframe.

1B: Ordinary Least Squares

Step 1: Basic Dataframe Operations
Perform the necessary steps to transform the dataframe in a format ready for machine learning.

Step 2: Perform Ordinary Least Squares
Perform Ordinary Least Squares (OLS) and answer the following question: what proportion of the outcome variable could be explained by the predictor, or feature, variables?

1C: Gradient Boosting Method

Step 1: XGBoost Machine Learning Algorithm
Create an XGBoost ML model in Python. Create a suitable variable name for the model.

Section #2: Explainable Machine Learning

2A: SHAP

Step 1: Create a SHAP Waterfall Plot of XGBoost ML
What are the Top 5 features that are correlated with the outcome variable?

Step 2: Create a SHAP Force Plot of XGBoost ML
Write no more than one paragraph summary of the insights shown by the SHAP force plot.

2B: Partial Dependence Plots (PDPs)

Step 1: Create a PDP of one feature against the outcome variable
Select one data feature and create a Python code that will display the PDP chart showing effects of this feature on the outcome.

Step 2: Select a second data feature and create a PDP against the outcome variable Select another, different, data feature and create a Python code that will display the PDP chart showing effects of this feature on the outcome.

Section #3: Predict Causal Factors

Step 1: Use EconML AI Causal Learner
Create a suitable variable name for the causal machine learner.
Complete the Python code to create an EconML causal learner.

Step 2: ATE Chart - Visualisation of Causal Factors
Write Python code that will display a chart of the Average Treatment Effect (ATE) magnitudes, or values, for all the data features.
Which of the data features are macroeconomic causal factors?
Given the macroeconomic causal factors, what recommendations would you make to the Australian government?

Important:

Complete the notebook within the allocated time of 3 hours during workshop.
Create a new "text/markdown" cell in Google Colab to record your answers. There's no need to create a separate report.
Email your version of the Google Colab Python notebook as a sharable link. Note: ensure that you create the correct "share" settings.